Chemical biology approaches for nuclear receptors : molecular … · • The final author version and the galley proof are versions of the publication after peer review. ... 4.2.

Chemical biology approaches for nuclear receptors :molecular and structural insightsCitation for published version (APA):Fuchs, S. (2013). Chemical biology approaches for nuclear receptors : molecular and structural insights.Technische Universiteit Eindhoven. https://doi.org/10.6100/IR750746

DOI:10.6100/IR750746

Document status and date:Published: 01/01/2013

Document Version:Publisher’s PDF, also known as Version of Record (includes final page, issue and volume numbers)

Please check the document version of this publication:

• A submitted manuscript is the version of the article upon submission and before peer-review. There can beimportant differences between the submitted version and the official published version of record. Peopleinterested in the research are advised to contact the author for the final version of the publication, or visit theDOI to the publisher's website.• The final author version and the galley proof are versions of the publication after peer review.• The final published version features the final layout of the paper including the volume, issue and pagenumbers.Link to publication

General rightsCopyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright ownersand it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights.

• Users may download and print one copy of any publication from the public portal for the purpose of private study or research. • You may not further distribute the material or use it for any profit-making activity or commercial gain • You may freely distribute the URL identifying the publication in the public portal.

If the publication is distributed under the terms of Article 25fa of the Dutch Copyright Act, indicated by the “Taverne” license above, pleasefollow below link for the End User Agreement:www.tue.nl/taverne

Take down policyIf you believe that this document breaches copyright please contact us at:[email protected] details and we will investigate your claim.

Download date: 10. Aug. 2020

https://doi.org/10.6100/IR750746

https://doi.org/10.6100/IR750746

https://research.tue.nl/en/publications/chemical-biology-approaches-for-nuclear-receptors--molecular-and-structural-insights(46146572-157a-49b1-b8f8-31945f74f718).html

Chemical Biology Approachesfor

Nuclear Receptors- Molecular and Structural Insights -

PROEFSCHRIFT

ter verkrijging van de graad van doctor aan de Technische Universiteit Eindhoven, op gezag van de rector magnificus, prof.dr.ir. C.J. van Duijn, voor een commissie aangewezen door het College voor Promoties in het openbaar te verdedigen op woensdag 27 februari 2013 om 16.00 uur

door

Sascha Fuchs

geboren te Gütersloh, Duitsland

Dit proefschrift is goedgekeurd door de promotor:

prof.dr.ir. L. Brunsveld

A catalogue record is available from the Eindhoven University of Technology Library ISBN: 978-94-6203-278-1

Cover Design: Maximilian Tomasoni & Sascha Fuchs Printing: Wöhrmann Print Service, Zutphen, the Netherlands

to my family

I

Table of content

CHAPTERONEMolecular Insights into Nuclear Receptor Structure and Function 11.1. Nuclear Receptor Superfamily 21.2. Nuclear Receptors in Chemical Biology 111.3. Nuclear Receptors as Drug Targets 141.4. Aim and outline of the thesis 201.5. References 22

CHAPTER TWOInvestigation into Posttranslational Modifications of the Estrogen Receptor Hinge Region 312.1. Introduction 322.2. Predicted helical character of model 12 mer peptides 342.3. Solid phase peptide synthesis of 12 mers 352.4. Structural analysis by far UV circular dichroism 362.5. Structural analysis using Nuclear Magnetic Resonance 372.6. Acetylation by lysine acetyltransferase p300 382.7. Discussion 402.8. Conclusion 432.9. Experimental 442.10. References 47

CHAPTER THREEStudying the Estrogen Receptor Hinge Region Using Expressed Protein Ligation 513.1. Introduction 523.2. Results and discussion 553.2.1. Generating the recombinant protein for EPL 553.2.2. Generating chemically synthesized thioester peptide for EPL 563.2.3. Test ligations using EPL with the model protein C YFP 583.2.4. N’ terminal elongation of the ER LBD towards the ER HR 613.2.5. Influence of hinge elongation on protein stability and coactivator binding 623.3. Conclusion 643.4. Experimental 653.5. References 68

II

CHAPTER FOURScreening the Estrogen Receptor Coactivator Interaction 714.1. Introduction 724.2. Ribosome display screening and preliminary validation 744.3. Exploration of the PXLXXLLXXP motif 774.4. Structural analysis of potent binders 794.5. Discussion 824.6. Conclusion 874.7. Experimental 884.8. References 97

CHAPTER FIVEIn Silico Design of Androgen Receptor Coactivator Binding Inhibitors 1015.1. Introduction 1015.2. In silico peptide design 1045.3. Molecular Dynamics Simulations – identification of scaffold 1055.4. First generation peptides – the second proline position 1075.5. Influence of first proline position and charge on structure 1075.6. Influence of the first proline position and charge on binding 1095.7. Discussion 1115.8. Conclusion 1165.9. Experimental 1175.10. References 121

CHAPTER SIXModulation of retinoid X receptor activity by biaryl natural products 1236.1. Introduction 1246.2. Initial screen of biaryl natural products isolated fromMagnolia species 1266.3. Concentration dependent effects of honokiol with and without agonist 1286.4. Insights into the honokiol RXR interaction 1286.5. Studies on the pharmacological profile of honokiol 1296.6. Modification of honokiol towards exclusive targeting of the RXR AF 2 1306.7. Discussion 1326.8. Conclusion 1356.9. Experimental 1366.10. References 140

ANNEXSummary IIIDeutsche Zusammenfassung für Nicht Biochemiker VIList of Publications IXCurriculum Vitae XAcknowledgements XI

CHAPTER ONE 4 Molecular Insights into Nuclear Receptor

Structure and Function

Abstract. Nuclear receptors (NRs) are multi domain transcription factors, typically undersmall molecule control. All proteins of the NR superfamily have a similar overallconformation and conserved domains. The ligand binding domain (LBD) receives smalllipophilic molecules via its buried ligand binding pocket (LBP). The LBD and the partlyinvolved hinge region (HR) are responsible for the subsequent dimerization of the receptorwhile the interaction with NR response elements is accomplished by the zinc fingers of theDNA binding domain (DBD) in cooperation with the HR. Rearrangement of theconformation of the LBD leads to the establishment of the activation function 2 (AF 2).Although this AF 2 is a main driving force for coactivator recruitment, transactivation of thereceptor is supported by activation function 1 (AF 1), located on the amino terminal domain(NTD). The available crystal structures of NRs have improved the molecular understandingof this complex ligand mediated mechanism of signal transduction. Structural and functionalanalysis has revealed that receptor specificity can be introduced at different points, such asthe size and shape of the LBP, size and structure of varying domains, including NTD and Cterminal F domain, and also by the appearance of the coregulators presenting motifs that areselectively accommodated by the AF 2. A consequence of the inherent conformationalflexibility of the AF 2 is also the ability of NRs to respond differently to the presence of anagonist, a partial agonist, an antagonist or a reverse agonist leading to the differentiatedrecruitment of coactivators, corepressors or coactivator binding inhibitors that ultimatelydetermine the transcriptional activation or repression of NRs.

Chapter One

2

1.1. Nuclear Receptor Superfamily

Members of the nuclear receptor (NR) superfamily are transcription factors that directlyinteract with DNA. Upon binding of small lipophilic molecules the structure of NRsundergoes a dynamic and conformational change, which leads to an enhanced ability torecruit other proteins, such as coregulators. This signal transduction is involved in severalphysiological functions across the metazoan kingdom, including development, homeostasis,reproduction, and metabolism.1,2 Although the term `hormones’ was first used in the earlystage of the 20th century3 not much was known then about their function until `a specifichigh affinity receptor’ was identified as the target for estradiol in the 1960s.4–6 This was thefirst biochemical evidence for the existence of NRs. However, the first cloned NR was theglucocorticoid receptor (GR, NR3C1), followed by the estrogen receptor (ER , NR3A1) inthe mid eighties.7,8 The full sequencing of the human genome finally identified 48 differentNRs,9,10 which can be divided into six evolutionary groups (subfamilies and exemplarymembers):11,12

Thyroid Receptors (NR1xx, thyroid hormone receptor like)o thyroid hormone receptors (TRs, NR1Ax)o retinoic acid receptors (RARs, NR1Bx)o peroxisome proliferator activated receptors (PPARs, NR1Cx)o RAR related orphan receptor (RORs, NR1Fx)

Retinoid Receptors (NR2xx, retinoid X receptor like)o hepatocyte nuclear factor 4 (HNF4s, NR2Ax)o retinoid X receptor (RXRs, NR2Bx)

Steroid Receptors (NR3xx, estrogen receptor like)o estrogen receptors (ERs, NR3Ax)o glucocorticoid receptor (GR, NR3C1)o androgen receptor (AR, NR3C4)

Nerve Growth Factor IB Receptors (NR4xx, nerve growth factor IB like)o nerve growth factor IB (NGFIB, NR4A1)

Steroidogenic Factor Receptors (NR5xx, steroidogenic factor like)o steroidogenic factor 1 (SF1, NR5A1)

Germ Cell Nuclear Factor Receptor (NR6A1)

Molecular Insights into Nuclear Receptor Structure and Function

3

All members of the NR superfamily share a common protein structure with highstructural homology and conserved domains:13 the variable amino terminal domain (NTD)including ligand independent activation function 1 (AF 1);14 the central highly conservedDNA binding domain (DBD) that can distinguish between nuclear receptor responseelements;15 the poorly conserved connecting hinge region (HR) featuring differentfunctionalities, including regulation of nuclear translocation;16 and the moderately conservedligand binding domain (LBD) that consists of 12 helices interacting with both ligands viathe ligand binding pocket (LBP), and coregulators via activation function 2 (AF 2).Additionally, a carboxy terminal F domain may or may not be present, and is attributed toplay a role in protein dimerization and the regulation of coactivator binding (Figure 1 1).17

Figure 1 1 | Domain structure of the nuclear receptor superfamilySchematic representation of the nuclear receptor superfamily: amino terminal domain (NTD),including activation function 1 (AF 1); DNA binding domain (DBD), including zinc fingerstructure; hinge region (HR); ligand binding domain (LBD), including ligand binding pocket(LBP), activation function 2 (AF 2) and binding function 3 (BF 3); and F domain. Left:representative domain structures of MR, AR, ER , RXR and VDR. Right: Crystal structure ofestradiol bound ER LBD interacting with proline based peptide including helix 1 – helix 12 (H1H12).

Chapter One

4

1.1.1.Amino terminal domain and activation function 1 (NTD/AF 1)

The amino terminal domain (NTD) is the most varying region of NRs in terms of both sizeand sequence. While the mineralocorticoid receptor (MR, NR3C2) exhibits a 602 residue longNTD, the vitamin D receptor (VDR, NR1I1) is only 23 amino acids long. The progesteronereceptor (PR, NR3C3) and GR with NTDs of similar size have only 15% sequence homology.This trend is widespread across the entire NR superfamily and even within subgroups suchas the steroid receptors.18 Consequently the capacity of the ligand independent activationfunction 1 (AF 1) to contribute to the constitutive activity of the receptor differssignificantly.19 Additionally, alternative splicing resulting in truncated receptor isoformsoccurs preferential to the exclusion of the N’ terminus affecting the NTD.20–22 Thedetermination of the tertiary structure of entire NRs is very complicated due to difficulties infull length NR expression and purification. While the structure of the intact PPAR :RXRdimeric complex could be solved (Figure 1 1), the NTD was not visible due to its highmobility. This view is coherent with studies into the dynamics of NR domains, whichrevealed the NTD to be intrinsically flexible.23 The conformation of the NTD in the context ofthe whole NR is currently described as a disordered domain with a partial, variable degree ofsecondary structure that converts to a stabilized structure as a consequence of both DNA andprotein protein interactions.24–26 Indeed, studies discovered structural changes in the NTD inpresence of the DBD or as a function of DNA binding for several NRs.27–30 Although thisintramolecular interaction does not lead to globular structure formation, circular dichroism(CD) and nuclear magnetic resonance (NMR) spectroscopy of the AF 1 (e.g. of GR) in thepresence of the helix stabilizing trifluoroethanol revealed the helix forming propensity ofthe NTD.31,32 Thus, the required secondary structure for AF 1 mediated activation is gainedupon interaction with other proteins, such as coactivators.33 This induced fit strategy ofincreased folding upon interaction with coactivator motifs could also be discovered withinother transcription factors, including p53 and VP16.34,35 Noticeably high numbers ofposttranslational modifications (PTMs) and their known influence on domain conformationand activity also suggests an important role for NTDs in NR function.18,36

The androgen receptor (AR, NR3C4) features a very unique transcriptional activationmechanism. In contrast to other NRs, where AF 1 makes only a minor contribution to itsoverall transcriptional activity, in AR the main part is facilitated by its AF 1.37 In line withother NRs, the secondary structure is also intrinsically unstructured and stabilized uponinteraction with other proteins.32 However, this interaction is not limited to intermolecularbinding partners, such as coactivators, but also includes binding to the C’ terminal LBD or,more specifically, the AF 2 via a binding motif based on bulky aromatic residues.38 Thisinterplay between AF 1 and AF 2 (N/C interaction) in AR is ligand dependent and isnecessary for AR mediated transactivation in vivo (Figure 1 2).39,40 To conclude, the active


5

conformation of AF 1 is induced by stabilization of the NTD by both intra (DNA, AF 2) andintermolecular (coregulator) interactions.

Figure 1 2 | N/C interaction in androgen receptor actionCurrent understanding of AR function: upon binding of the ligand to the ligand binding domain(LBD) of AR, rearrangement of 12 (H12) facilitates intramolecular interaction between theN’ terminal domain (NTD) and the activation function 2 (AF 2) via a FXXLF motif. Subsequentnucleolar translocalization leads to binding to the DNA and recruitment of coactivator featuring aLXXLL motif.

1.1.2.DNA binding domain (DBD)

The first tertiary structure of a DNA binding domain (DBD) in complex with DNA wassolved for the GR by nuclear magnetic resonance (NMR) and X ray crystallography in theearly nineties.41,42 Later, other DBDs were described with and without the DNA responseelement.43,44 This domain of approximately 70 residues is highly conserved across NRs andexpresses a globular fold containing two zinc ions (zinc finger structure), which arenecessary for the conformation and activity of the domain.45 Two helices are present: one(helix 1) interacts base specific with the major groove of the DNA and the other (helix 2)stabilizes this complex. Upon binding of one monomer a conformational change in the DBDfacilitates the dimerization of a second monomer.46,47 These two monomers can be either thesame type of NR, resulting in homodimers that bind to palindromic inverted repeats on theDNA (steroid receptors), or RXR forming heterodimers with other NRs binding to directrepeats on the DNA (PPAR, VDR among others). However, RXR can also form homodimers,with the conformation of the less conserved C’ terminal extension (CTE) of the DBD

Chapter One

6

extending into the hinge region (HR) playing an important role in the decision to formhetero or homodimers.48 In that sense, the CTE can either provide an additionaldimerization interface to enhance interaction with RXR,47,49 or play an additional role in theinteraction of the minor groove of the DNA (compare Figure 1 7, PPAR HR),23 similar toorphan receptors that bind only as monomers to extended response elements.50 Recently, thestructure of intact PPAR RXR (both full length) in complex with DNA could be solved(Figure 1 7), which revealed the induction of DNA interaction of both DBDs by the PPARLBD and is in accordance with findings that demonstrated reduced DNA binding uponmutations in the LBD of PPAR .23

1.1.3.Hinge Region (HR)

The hinge region (HR) of NRs is much less highly conserved compared to the flankingDBD and LBD. Apart from its flexibility, which enables different orientations of the domainsof NRs with respect to each other the presence of the nuclear localization signal (NLS) wasalso discovered early on to be at least partially located in the HR of NRs.2,11 Indeed, theorientation between the NTD, including ligand independent AF 1, and LBD, which includesthe ligand dependent AF 2, is based on the composition and length of the HR and thereforeinfluences transactivation.51 Further studies revealed the involvement of HR as amultifunctional region in dimerization,15 nuclear translocation,52,53 DNA binding,54–56

transactivation,16,57,58 and receptor mobility.59

Similar to the NTD, the HR is highly decorated with residues accessible for PTMs.Modifications such as acetylation are reported to be connected with the folding of AR,resulting in the regulation of transcriptional activity,57,60,61 subcellular distribution, andcoactivator and corepressor binding, respectively.16,58,60,61 Similar discoveries were made forER.62 Both small modifications such as phosphorylation, acetylation or methylation atvarious positions63–65 and the interaction of coregulators66 are known to stabilize the receptorleading to enhanced transcriptional activity. Simultaneously, the HR is a target for significantchanges, such as mono or polyubiquitination, resulting in proteosomal degradation.67,68

Interestingly, some phosphorylation sites, including S305 and T311, are connected to thefailure of tamoxifen treatment in ER positive breast cancer. However, S282, among others, ispredicted to improve tamoxifen responsiveness.69,70

The mobility of RXR HR enables receptor interaction with various response elements asheterodimers.71,72 However, the interaction with these heterodimer partners is influenced bythe HR of the interaction partners but not by the RXR HR itself. The interaction with NRcorepressors was also discovered for the RXR heterodimer partner TR, but not for RXRitself.73,74 The intramolecular interplay between HR and LBD is confirmed by the release ofthese corepressors upon ligand binding to the LBD.


7

While the HRs of both members of the PPAR RXR heterodimer pair are rather flexible, thestructurally similar HRs of two other RXR interaction partners TR and VDR, respectivelyare unique due to their helical structure as an extension of the DBD.47,75,76 While the HR wasnot or only partially visible in crystallization analysis even in intact PPAR RXR incomplex with DNA23 cryo electron microscopy of VDR RXR allowed for the positioning ofcrystal structures of LBD and DBD, and for the first time the visualization of the HRs of bothreceptors.77 The crystal structure identifies that a conformationally well defined VDR HR isimportant for the stabilization of the open conformation that does not require any interactionbetween LBD and DBD. Mutational studies revealed that the conserved HR length ratherthan the sequence of VDR is important, which does not apply for RXR.78 This validates thehypothesis for the HR of VDR and TR that upon DNA interaction the HR is repositioned,thereby modulating the dimer interface and consequently transactivation.76,79,80

Taken together, the HR is poorly conserved across all members of the NR superfamily.This then provides a platform for possible specificity between proteins that are highly ormoderately conserved at important regions including DNA, ligand, and coregulatorinteraction.81

1.1.4.Ligand binding domain and activation function 2 (LBD/AF 2)

In the mid nineties the first structures of NR ligand binding domains (LBDs) were solvedin the ligand free apo form (RXR LBD)82 and the more compact ligand bound holo form(RAR LBD and TR LBD), respectively.83,84 To date, structures are available for nearly allsubfamilies of NRs that feature a canonical fold including 12 helices (H1 H12) orientated ina three layered anti parallel fashion. Mutation and deletion studies of various LBDsdiscovered the most C’ terminal helix 12 (H12) being of particular importance for liganddependent activation of the receptor.85–87 Indeed, the apo form identified H12 projecting awayfrom the structure,88,82 however, upon ligand binding rearrangement of H12 occurs in theform of a folding over of the core structure of the domain and closing of the LBP.82,89–93 In thisholo form, H12 and residues from H3 H5 generate the activation function 2 (AF 2) whichconsists of a hydrophobic groove limited in size by two charged amino acids in H3 and H12(charge clamp).94,95 The AF 2 was identified as a target for several coregulators which interactvia a leucine rich LXXLL binding motif (L, leucine; X, any amino acid).96 Furtherinvestigations revealed this motif embedded in a well defined two turn helix to benecessary, but also sufficient for NR coactivator interaction.97

Ligands that bind to the LBP of the receptor regulate this protein protein interaction bymeans of a repositioning of the H12. Stabilization of this helix can be reached by directcontact with the ligand,90 a stabilizing effect on helices close to the LBP by the ligand,92 orthrough long range interactions between LBP and AF 2.98 Ligand selectivity is achievedthrough cooperative function of H3, H5, and H11, which feature residues capable of

Chapter One

8

hydrophobic interactions and specific hydrogen bonding networks. The size of the LBPburied in the LBD can vary between different members of NRs between 450 and 1,600 Å3.99

Apart from receptor activating ligands (agonists), receptor inhibiting ligands (antagonists)can also target the LBP. Upon binding of an antagonist ligand to the LBP, H12 is eithersterically hindered to transition into the holo form,100,101 or instead covers the AF 2 resulting inan active blockage of the binding of coactivators.102 Another way to silencing receptor activityis through the recruitment of corepressors, which takes place in ligand free nonsteroidalreceptors, such as thyroid and retinoid receptors. This protein protein interaction is mediatedvia an extended LXXLL like motif that binds to the AF 2. In this way, the lack of H12 resultsin the ability to accommodate the one turn enlarged helices – including sequences such asLXXI or HIXXXL/I, respectively exposed by corepressors.73,74,103,104,105 This NR corepressorcomplex can be stabilized by antagonists since they can provide an even more open AF 2than is present in the apo form. Indeed, even steroid receptors – except AR,106,107 whichnaturally does not bind to corepressors108 – were discovered in a DNA corepressor complexrepressing transcriptional activity.109,110 In addition to these H12 related means of modulatingNR activity, RXR regulation can be accomplished by steric hindrance of AF 2. The formationof tetramers renders coactivator binding impossible, whereas upon ligand binding thereceptor exists in the active dimer state.111

As previously discussed, AR possesses a unique mechanism through which transcriptionalactivity can be modulated. The intramolecular interaction between the unstructured AF 1 inthe NTD and the ligand dependent AF 2 in the LBD (N/C interaction) is required for therecruitment of coactivators (Figure 1 2).37,112,113 A phenylalanine rich motif (FXXLF) on theNTD facilitates the interaction with the AR AF 2 by providing more space on the proteinsurface – coated with less bulky residues compared to other NRs.114–116 Further studiesfocusing on the atypical NR binding groove revealed that other bulky aromatic residues areaccommodated, while the NR binding motif LXXLL showed less binding affinity comparedto other receptors.117–120

Recently, another interaction platform on the LBD was discovered in AR: the bindingfunction 3 (BF 3), which is located adjacent to AF 2 and involves principally H9 and partlyH1.121 Structural and functional studies suggested a possible allosteric function for BF 3(modulating AF 2) transmitted through amino acid residues adjacent to AF 2.122 While nonatural target has yet to be directly identified,123 studies on FKBP52 induced AR activityrevealed the BF 3 surface as a putative FKBP52 interaction and regulatory surface.124

Superimposition of the solved crystal structures of other NR LBDs identified similarsequence and structural identity, presumably in steroid receptors but also in othersubclasses.121

Apart from ligand and coregulator binding and thus signal transduction, the LBD of NRsalso features a dimerization surface, with H9 and H10 being the main helices involved.


9

Hydrophobic amino acids with flanking charged residues constitute a cluster for interactionwith other LBDs. Although the general principle is the same, certain differences could bediscovered between heterodimer and homodimers formation, for instance the involvement ofH11 and H12 in the case of the PR.125

1.1.5.F domain

The F domain is located at the very C’ terminus of NRs, but is not present in allreceptors.24,126 Where present, the F domain varies in size 19 to 80 residues and is poorlyconserved. Indeed, between the ER subtypes and the F domain varies by ten amino acidsin length and with less than 25% homology.127,128 However, this variation and itsinvolvement in both receptor dimerization and coregulator recruitment makes it partiallyresponsible for the different activities discovered for the ER subtypes.17,129 131 Although thedomain is required for the agonistic effect of tamoxifen on the ER element,132,133 it is notnecessary for ligand dependent transactivation, and can even enhance it.24 While there iscurrently no structural data available for ER, other NRs have given (partial) structuralinsights: visible as an extended strand in PR;134 involved with dimer partners in RAR.135,136

Investigations on the hepatocyte nuclear factor 4 (HNF4 ) also confirmed the importance ofthis domain. With 60 residues, it is the longest F domain known among NRs and isresponsible for different transcription activities (four fold) in two HNF4 isoforms.137–140

1.1.6.Posttranslational Modifications

Posttranslational modifications (PTMs) play an important role in NR function.141 Thesereceptor modifications can be achieved by either the transfer of a functional chemical group(e.g. phosphate and acetyl) or the coupling of other polypeptides (e.g. sumoylation andubiquitination). Studies have connected various PTMs to several diseases, withphosphorylation, acetylation and sumoylation of AR, ER , GR and PPAR being the mostreported kinds.For decades now, advanced prostate cancer has been successfully treated through hormone

deprivation therapy. However, increased death rates have been linked to hormone refractoryprostate cancer (HRPC) as a function of AR phosphorylation.142 Interestingly, the samemodification on the same residue leads to reduced toxicity of the elongated polyglutamine inKennedy’s Disease.143 Acetylation of AR is also connected to both an increase in cell growthin prostate cancer tumor models and antagonist resistance.58 Furthermore, sumoylation ofAR is directly linked to the onset of prostate cancer.144

In the case of ER positive breast cancer, some posttranslational phosphorylation events,including S118 and S167, are linked with an improved response to tamoxifen and aromataseinhibitor treatment,145–147 while others, such as S305, T311 and S559, are associated with apoorer response.69 Analysis of breast cancer samples identified a predominantly high lysineto arginine mutations rate in ER leading to a reduced susceptibility to acetylation. Since this

Chapter One

10

mutation also results in enhanced estradiol sensitivity, acetylations were attributed to thesuppression of this ligand sensitivity and thus a lowering of the risk of breast cancerdevelopment.148–151

Glucocorticoid resistance in the treatment of various inflammatory diseases is a result ofboth GR phosphorylation and acetylation.152–155 While posttranslational phosphorylation ofPPAR is also linked to insulin resistance and possibly to obesity156 Furthermore, PPAR hasanti inflammatory effects by sumoylation mediated recruitment of NR corepressor, resultingin the silencing macrophage inflammatory genes.157–159 In summary, PTMs are anothercomponent in the regulation of NR function, which can be used as to measure diseasedevelopment as biomarkers that provide information about the patient responsiveness toNR mediated drug therapy.141


11

1.2. Nuclear Receptors in Chemical Biology

The basic function of NRs are closely related to their conserved structure, especially of thetwo domains DBD and LBD. These are mainly involved in communicating with thesurrounding environment,13 such as small molecules interacting with the buried ligandbinding pocket,89–93 DNA binding to Zn finger structure,45 and coactivator proteins binding tohydrophobic groove.94,95 This phenomenon of cellular response controlled by small lipophilicmolecules is the ideal platform for chemical biology approaches using the receptor interfaceswith ligand, DNA, and coregulators as targets for the design of biochemical tools.160

1.2.1.Protein degradation

Induced protein degradation is one possibility to inhibit NR function. The design of theproteolysis targeting chimeric molecule (PROTAC) is a posttranslational chemical genetictechnique to induce protein degradation by means of the ubiquitin proteasome system.161 ThePROTAC consists of both the E3 ubiquitin ligase recognition motif and a recognition motiffor the target protein.162,163 Until now, PROTACS targeting AR and ER have been developedthrough chemical labeling of the endogenous ligands dihydroxytestosterone and17 estradiol (Figure 1 3).164,165 Another approach uses a three hybrid small molecule proteininteraction system in which split ubiquitin functions as a sensor. In a model system using thetwo ligand receptor pairs GR/dexamethasone (Dex) and dihydrofolate/methotrexate (Mtx),two halves of ubiquitin could be successful reconstituted. In this case, a small bivalent hybridmolecule Dex Mtx functioned as the dimerizer for dihydrofolate and GR resulting inubiquitin specific protease cleavage and subsequent degradation.166

Figure 1 3 | NR specific PROTAC used for the induced degradation of NRs

1.2.2. Intein mediated protein engineering

Protein splicing is a self catalyzed posttranslational process resulting in the excision of aninternal protein segment (intein).167 This intein mediated self cleavage activity is a powerfultool to activate the protein of interest in a controlled fashion. One approach is expressedprotein ligation (EPL) in which a peptide featuring a N’ terminal cysteine is ligated to arecombinantly expressed protein bearing a C’ terminal thioester.168 This thioester can begenerated by the fusion protein consisting of protein and C’ terminal intein.169 EPL was usedto investigate the dynamic mechanism of NR activation by incorporation of a fluorescent

Chapter One

12

label in the relevant H12 of the PPAR LBD. Anisotropy measurements revealed a reducedmobility of H12 upon ligand binding. Thus, these findings were the first direct evidence forthe molecular switch of H12 being essential for transcriptional activity.170 The same approachwas used in a study concerning the influence of tyrosine phosphorylation on H12 of ER .Crystal structure analysis of the site specific posttranslationally modified protein gaveinsights into structural conformation as a function of H12 phosphorylation.171

Another way of modulating protein activity is through the integration of inactive splitinteins that are activated by small molecules.172 Expression of both a split intein and a splitgreen fluorescent protein (GFP) separated by the VDR LBD resulted in a intein basedvitamin D fluorescent biosensor leading to detection of ,25 hydroxyvitamin D3 in thepM range. In this case, ligand binding induces a conformational change in the LBD, whichbrings the two intein halves together with subsequent protein splicing and reconstitution ofGFP the read out for the binding event (Figure 1 4).173 A similar approach for the detection ofNR agonists used a modified galactosidase assay based on both ER LBD and a splitintein. Upon agonist binding the protein hydrolyzes the glycosidic bond resulting inyellow O nitrophenol.174

Figure 1 4 | Split inteins as a tool in agonist induced reconstitution of fluorescence

1.2.3.Photo caged ligands

Photo caged bioactive substrates, which are biologically inert small molecules that gaintheir function upon exposure to light, can function in a unique spatial and temporalmanner.175 The introduction of photo caged ligands for NRs allows for light activated geneexpression (LAGE), such as via 3’ hydroxyl inactivated estradiol, which activates ER uponirradiation with UV light (Figure 1 5).176 In a reverse approach, selective receptor modulators(SERMs) have also been photo caged through the introduction of the same nitroveratrylgroup at the hydroxyl group of the SERM.177 A similar design has resulted in the lightdependent targeting of other NRs, such as RAR , TR , and VDR.160,176

Figure 1 5 | Photo caged ligands for NRs


13

1.2.4.Genetically encoded sensors

The generation of genetically encoded sensors is another possibility for the investigation ofcellular processes.178 The introduction of fluorescent proteins allows the use of techniquessuch as Förster resonance energy transfer (FRET), fluorescence recovery after photobleaching(FRAP) or Fluorescence lifetime imaging microscopy (FLIM).179–181 In the NR field thesefluorescence based approaches have been widely used. Umezawa et al. used a fluorescentindicator to discriminate between estrogen receptor agonists and antagonist in living cells. Aconstruct consisting of the FRET donor acceptor pair cyan/yellow fluorescent protein(CFP/YFP) was separated by a sequence based on the NR coactivator 1 (NCOA1/LXXLL)fused to the N’ terminus of ER LBD. Agonist binding led to recruitment of the LXXLL motifto the AF 2, leading to close proximity of the FRET pair and thus FRET. However, in theantagonist bound conformation both domains are held apart and no FRET can occur (Figure1 6).182,183 A reverse approach was used in the development of a bile acid sensor for real timeintracellular imaging in single living cells, consisting of bile acid binding FXR an LXXLLbased peptide and a FRET pair. Mutations in the domains of the cerulean/citrine FRET pairled to intramolecular association and thus efficient FRET in the ligand free state. However,bile acid induced FXR LXXLL interaction resulted in reduced FRET signal as a function ofincreased distance between the FRET pair.184 In a similar way the conformational change ofAR and its N/C interaction was used to monitor subcellular distribution and ligand inducedconformational change. However, in this particular case the fusion protein consists only offull length AR with one component of the CFP YFP FRET pair on each side. Liganddependent N/C interaction and the induced proximity of FRET pairs resulted in FRET.184,185

Houtsmuller and coworker took this approach one step further and performed simultaneousFRET and FRAP measurements to study the spatiotemporal organization of AR in livingcells. They discovered that the N/C interaction is a function of AR mobility. While present insolution, upon binding to DNA, this interaction is abolished to provide space for coactivatorinteraction.186

Figure 1 6 | FRET based biosensing of NR agonists

Chapter One

14

1.3. Nuclear Receptors as Drug Targets

The superfamily of nuclear receptors (NRs) is involved in a wide range of physiologic butalso pathophysiologic processes.11,187 Already in the early stages of the 20th century it wasdiscovered that both ovariectomy decreases chance of mammary cancer and extracts fromadrenal gland can be used in the treatment of Addison’s disease based on glucocorticoiddeficiency.188,189 Purified cortisone utilized against inflammatory diseases and itssubsequent total synthesis (1948) was pioneering work for subsequent syntheses of steroidcompounds.190,191 In the 1970s tamoxifen was identified as an anti estrogen in the screen ofnonsteroidal compound libraries and became a leading drug, mainly in the treatment of earlybreast cancer, but also in ductal carcinoma, and breast cancer chemo prevention.189,192 195 Theability to control NR function through the binding of small lipophilic molecules has madeNRs one of the most targeted classes of proteins in the field of drug discovery. Indeed,statistical analysis of marketed drugs revealed that over 50% interfere with only four proteinclasses, including G protein coupled receptors, NRs (13%), ligand gated ion channels orvoltage gated ion channels, with GR and the histamine H1 receptor being the top targets.196,197

Diabetes (ER, PPAR), asthma (GR), atherosclerosis (LXR), osteoporosis (ER, VDR), cancer(ER, AR, GR, RXR), arthritis (GR) and heart failure (MR) are only some examples of manydiseases currently treated with drugs that mainly target the ligand binding pocket (LBP) ofNRs.197,198 However, aside from the LBP, other alternative sites of NR modulation are underinvestigation, including the activation function 2 (AF 2), binding function 3 (BF 3), the zincfinger motif, and the NR response element (Figure 1 7).123


15

Figure 1 7 | Targeting nuclear receptorsOutline of different types of small molecule modulators of NRs function. Six groups of targetsshow either natural binding partners or small molecule inhibitors as a result of de novo design ordiscovered from high throughput screening. The center of the figure shows the crystal structureshowing both intact PPARy (DBD, HR, LBD) and RXR (DBD, LBD) bound to DNA togetherwith buried ligands and LXXLL based coactivators (PDB, 3DZU).23

1.3.1.Targeting the Ligand Binding Pocket

The interaction of ligands at the NR ligand binding pocket (LBP) results in aconformational change of the LBD, which produces an interface for coactivator binding(AF 2). This activation or agonism of NRs by small lipophilic molecules has inspired thedesign of compounds that block this function (Figure 1 8, small molecule). For the first timeantagonism was structurally confirmed with the crystal structure of ER LBD bound toeither 4 hydroxytamoxifen or raloxifen.102,199 Although the antagonist induced overallstructure is similar to the agonist bound form (H1 H11), either bulky side chains of thechemical molecules lead to incomplete/alternative folding of H12 (blocking AF 2) or it is notat all part of the AF 2 (incomplete AF 2).92 Either way, coactivators are not recruited to AF 2.However, since corepressor and coactivator share parts of the binding site, antagonist canalso lead to reduced NR corepressor interaction as discovered for RAR bound to BMS614.200

Chapter One

16

Total inhibition of NR function, including their basal transcriptional activity, is aconsequence of the interaction of inverse agonists. In this case H12 is stabilized in a differentposition that allows the binding of corepressor.101 Inverse agonists of RAR that can naturallyrecruit corepressors significantly induces these silencing interactions.200,201 However, evenNRs that do not recruit corepressors innately, are capable of changing their conformationupon inverse agonist binding in a way that allows corepressors to bind.110,202

Partial agonists, another class of ligand, are poorly understood, as their NR bindingmechanism differs from those of agonist and antagonist. The functionality of partial agonistsdepends on the promoter context and cellular concentration, as shown for PPAR .203 Similarto agonists, they induce a confirmation that allows the recruitment of coactivators, however,poor H12 stabilization leads to a receptor conformation that is still capable to interact withcorepressors. Studies on LXR confirmed the hypothesis that the tissue specific coactivatorcorepressor ratio determines the activity of partial agonists.204

Selective NR modulators are similar to partial agonists in terms of their differentfunctionalities in various tissues. However, they not only show different levels of agonism,but even antagonistic effects. The first discovered selective ER modulator (SERM) was4 hydroxytamoxifen, initially thought to be an exclusive antagonist.109 Functioning as anantagonist in mammary cells, it functions as a partial agonist in endometrial cells.205 Likepartial agonists the presence of coactivators and corepressors are responsible for the SERMagonistic or antagonistic activity, respectively.206 Apart from other ER targeting SERMssimilar molecules for AR, PPAR and PR have also been discovered.207–210

1.3.2.Targeting the activation function 2

An alternative approach to classical ligands is the direct targeting of the NR coactivatorinteraction (e.g. AF 2). Due to the large number of crystal structures available for nearly allnuclear receptors and the detailed characterization of this protein protein interaction, it hasgained strong attention in the field of drug discovery.94 Overcoming the limitations of LBPbinding therapeutics, including insufficient selectivity and drug resistance due to long termtreatment, were the initial driving force to explore other druggable interactions on the NRsurface.123 Due to the limited length of the helix of NR coactivators two distinct residueson both sides of the AF 2 – it is unique among protein protein interaction interfaces that arepreferably greater in size.211,212 Thus, the design of coactivator binding inhibitors (CBIs) is apromising strategy to block NR transactivation.


17

Figure 1 8 | Targeting the NR ligand binding domainSchematic representation of targeting the ligand binding domain (LBD) of nuclear receptors(NRs): either by antagonist competing with the ligand for binding to the ligand binding pocket(LBP) or by coactivator binding inhibitors (CBIs) as peptidomimetics or peptide based inhibitorsbinding the ligand bound LBD at the activation function 2 (AF 2).

The design of small molecules that mimic the well structured helix of coactivators(Figure 1 8, peptidomimetics) combines the high selectivity of natural peptides with thebioavailability of small lipophilic drug like compounds.213 Another advantage compared topeptide based approaches is the high structural diversity since there are no limitationsplaced on the choice of side chain residues. However, the optimal arrangement of the sidechain mimics achieved in nature through the rigidity of the helix is challenging toimitate.214 In a de novo design based on a pyrimidine core the leucine side chains of theLXXLL motif were successful mimicked using branched alkyl substituents, which resulted inan ER selective coactivator binding inhibitor (CBI).215 The introduction of aromatic benzylicgroups as a mimic for the phenylalanine rich motif (FXXLF) involved in AR AF 2 bindingled to AR selective CBIs.216 Other attempts were based on biaryl scaffolds enabling out ofplane projection of side chain residues mimicking residues i, i + 3 to prevent stericrepulsion.217 A similar core structure was used to specifically target the charge clamp, whichemerges from two charged residues at both ends of the AF 2. In addition to the branchedalkyl side chains, substituents were introduced to address electronic interactions with thecharged amino acids.218 Another way to identify lead compounds that interact with a certaintarget is the high throughput screening (HTS) of large compound libraries. The combinationof an automated fluorescence based assay with X ray soaking studies discovered several leadcompounds targeting the AR. The associated structural analysis directly gave insights intothe location and mechanism of receptor interaction and revealed a BF 3 binding site adjacentto the AF 2.122 Another FRET based approach involved the screening of an 86,000 stronglibrary with diverse compounds. Subsequent structure activity relationship (SAR) studiesand cell based assays identified several CBIs.219,220 Modeling studies, however, revealed amuch reduced interaction with the receptor surface compared to model peptides based oncoactivator sequences. These findings suggest a possible non perfect fit of these CBIs,

Chapter One

18

resulting in an entropically unfavored state and might provide a structural explanation forthe difficulty of optimization M lead structures towards higher nM potency.219 Covalentinhibitors of TR have also been developed which are small molecule based butmechanistically different from the classical CBIs discussed here. In this case, a cysteine in theAF 2 of the NR reacts with the aminoketone of the inhibitor to form an irreversible Michaeladduct resulting in inhibition of coactivator recruitment.221–223

Difficulties in the optimization of small molecule as CBIs due to an imperfect fit with thereceptor AF 2 has (re)stimulated the design of peptide based CBIs (Figure 1 8, peptide basedinhibitor). Phage display is a powerful technology to identify the optimal peptide sequencefor a specific target.224–226 Combinatorial phage display against ER revealed three differentclasses of peptides with CBI functionality. In some cases NR selectivity could be achieved bythe context of the flanking residues of the common LXXLL motif.97 Screening the AR surfacewith phage display discovered, aside from the FXXLF motif, other aromatic residues capableof binding the AR AF 2. Structural analysis showed a perfect match of the identified hits andthe AR surface. Additionally it explained the preferential binding of aromatic based peptidescompared to LXXLL, which forms only a single hydrogen bond between the backbone andthe AR charge clamp instead of two in other NRs.227 Targeting other NRs in a similar wayrevealed different enlarged consensus motifs, such as HPLLXXLL in the case of PPAR andMPXLXXLL for MR.228,229 Another study also used phage display screening to obtaininformation about the influence of LXXLL flanking residues on ER binding. Due to thelocalization on well defined helices on miniproteins, this way of screening identifiesresidues that are optimal for protein binding, however, disconnecting their effect on helixstability.230,231

Next to low cell permeability and weak cellular stability, stabilization of the isolatedhelix is one of the major challenges on the way to drug like structures.232 Thus, stabilizing

the helix, which contains residues targeting the AF 2 (LXXLL/FXXLF), has been the goal ofvarious studies. The connection of two side chains by both a i, i + 4 lactam bridge or i, i + 3disulfide cyclization showed helical character, effective binding, and receptor selectivity ofthese small LXXLL peptides.233 In a similar approach using computational design andcombinatorial chemistry both ERs and TR could be selectively targeted. Based on thedocking studies, the leucines were sequentially substituted with hydrophobic non naturalside chains and additionally stabilized by a lactam bridge.234,235 In a separate study, therigidity of the peptide backbones was enhanced by the introduction of a hydrocarbon stapleat i, i + 3, i, i + 4, or i, i + 7. Apart from hydrophobic interactions of the leucine side chain,structural analysis identified additional van der Waals contact of the staple contributing tothe ability of the peptide to bind ER.236


19

1.3.3.Targeting alternative sites

Crystallographic HTS against AR identified aside from CBIs targeting the AF 2compounds targeting an alternative site on the LBD.122 This binding site, called bindingfunction 3 (BF 3), is located in close proximity to the AF 2, and is clearly distinguishable. Intotal seven molecules in the mid M range were co crystallized with the AR LBD, whichrevealed the recognition of residues that form part of both BF 3 and AF 2 and are thereforeallosterically disruptive of coactivator recruitment (Figure 1 7, BF 3).237 Another approachfocused on the identification of inhibitors for the positive AR regulator FKBP52. Themodified receptor mediated reporter assay was used to screen a natural compound library,out of which two compound candidates were discovered that inhibit FKBP52 enhancedmutated AR P723S function. Interestingly the responsible binding motif on the AR surfaceoverlaps with the recently discovered BF 3 function.124 Inspired by this finding, an in silicoscreen of AR BF 3 identified several BF 3 targeting ligands. Structural analysis frequentlyidentified dual binding properties of the active compounds, namely binding to both BF 3 andAF 2.238

Other work focused on targeting of the nuclear response element of NRs. The high affinitymolecular recognition of the minor groove by pyrrole imidazole polyamides led to ablocking of both ER and ERR (Figure 1 7, response element).239–241 A similar strategy wasused to target the AR element. Hairpin and cyclic polyamides showed a high ability toreduce prostate specific antigen (PSA) mRNA levels in androgen sensitive cancer cells(LNCaP).242,243 Another study concerning the treatment of ER positive human breast canceralso focused on DNA, however, targeting directly the zinc finger of the DBD. Theelectrophilic disulfide benzamide (DIBA) inhibits cell growth of tamoxifen resistant breastcancer in vivo (Figure 1 7, zinc finger).244,245 Recent studies focused on targeting the NTD ofAR. Both a small biaryl based molecule and chlorinated peptides could be discovered toinhibit transactivation of the N’ terminus of AR in prostate cancer cells.248–250

HTS also identified compounds with high binding affinities for AR. The molecularinteraction mechanism, however, is unclear. Two hits were identified that were synergisticwith each other suggesting different binding sites and were neither competing with theendogenous ligand nor binding to the AF 2. Furthermore, one hit showed improvedinhibition of LNCaP cancer cell proliferation and, in addition to its anti androgen function,caused a reduction in prostate weight (Figure 1 7, unknown).184,185 In a mammalian twohybrid assay, two other compounds were discovered that showed a similar profile. Norecruitment of AR to the PSA enhancer in combination with a lack of influence on subcellularlocalization suggested the interference at the level of DNA or coactivator binding.246,247

Chapter One

20

1.4. Aim and outline of the thesis

The transcriptional activity of nuclear receptors (NRs) involves a complex interplaybetween, on the one hand the conserved DNA binding domain (DBD) and the ligand bindingdomain (LBD), and on the other hand the less well conserved amino terminal domain (NTD),the hinge region (HR) and the carboxy terminal F domain. Major structural elementsinvolved in this activation event are activation function 1 (AF 1) at the NTD, the zinc fingermotif on the DBD, and ligand binding pocket (LBP) and activation function 2 (AF 2) bothlocated at the LBD. Perturbations in this highly sensitive regulatory machinery can lead tovarious disease states. Modulation of this permanently activated process can be achieved atdifferent positions, such as the LBP or AF 2.The aim of the thesis is to gain some molecular and structural insights into NR function

with a main focus on both the hinge region (HR) of the estrogen receptor (ER) and theactivation function 2 (AF 2) of ER, androgen receptor (AR) and retinoic X receptor (RXR).These structural domains and their role in protein protein interactions are also investigatedas possible drug targets for small lipophilic molecules and peptide based binders,respectively.In chapter 2 the structural influence of the acetylation of lysines 266 and 268 located in the

estrogen receptor (ER ) hinge region (HR) are investigated. Molecular dynamics (MD)simulations, circular dichroism (CD) studies and nuclear magnetic resonance (NMR)spectroscopy provided information about the change in secondary structure. Furthermorethe ability of the acetyltransferase p300 to acetylate these lysines was characterized.Chapter 3 deals with the semi synthesis of ER LBD combined with the HR. The

site specific introduction of posttranslational modifications (PTMs) was achieved by meansof expressed protein ligation (EPL). Generating both recombinant ER LBD with anN’ terminal cysteine and peptides with C’ terminal thioester were the requirements toperform the native chemical ligation (NCL). Furthermore the progressive extension of theER LBD in the direction of the HR forms part of a second study to acquire some insightsinto the function of this short and poorly conserved region of the receptor.In chapter 4 ribosome display was used to effectively screen the surface of the estrogen

receptor (ER) ligand binding domain (LBD) for novel natural peptide binders. Validated byin vitro and in vivo methods, structural insights were achieved by molecular dynamics (MD)simulation and X ray crystallography studies. Optimized proline based peptides were testedand compared to natural binding sequences in subsequent binding polarization and cellularassays. Co crystallization with ER LBD revealed a proline mediated binding mechanismbased on helix stabilization.


21

Chapter 5 describes an in silico design approach to target the AR. Design of the peptideswas based on separate results from ER screening. MD simulation studies were performed togain information about the propensities of the designed peptides to form helices. Thesemodeling studies were used to direct the solid phase synthesis of selected peptides. CDstudies were performed to confirm the predicted helical character. The AR bindingaffinities of the peptides were determined by means of a competitive polarization assay andcellular studies.In chapter 6 NR coregulator interaction profiling was performed to test a small compound

library of biaryl natural products (NPs) for potential RXR binding. This microarray basedtechnique allowed for a rapid and simultaneous screen of these molecules against knowncoactivator and corepressor proteins important for NRs. Structure activity relationshipstrends were assigned and the best binders verified in a polarization assay format.Gaining molecular and structural insights to understand function and regulation of the

structural elements of NRs is not only important to understand receptor pharmacology, butalso to provide novel entries to target these proteins.

Figure 1 9 | Outline of the thesis

Chapter One

22

1.5. References

1. G. I. Owen and A. Zelent, CMLS, Cell Mol Life Sci, 2000, 57, 809–827.2. V. Laudet and H. Gronemeyer, The Nuclear Receptor FactsBook, Academic Press, 1st edn., 2001.3. E. Starling, The Lancet, 1905, 166, 339–341.4. E. V. Jensen and H. I. Jacobson, in Biological Activities of Steroids in Relation to Cancer, Academic

Press, NY, USA, 1960, 161–174.5. J. I. Macgregor and V. C. Jordan, Pharmacol Rev, 1998, 50, 151–196.6. E. V. Jensen, Nature Medicine, 2004, 10, 1018–1021.7. S. M. Hollenberg, C. Weinberger, E. S. Ong, G. Cerelli, A. Oro, R. Lebo, E. B. Thompson, M. G.

Rosenfeld, and R. M. Evans, Nature, 1985, 318, 635–641.8. S. Green, P. Walter, V. Kumar, A. Krust, J. M. Bornert, P. Argos, and P. Chambon, Nature, 1986,

320, 134–139.9. E. S. Lander, L. M. Linton, B. Birren, C. Nusbaum, M. C. Zody, et al., Nature, 2001, 409, 860–921.10. J. C. Venter, M. D. Adams, E. W. Myers, P. W. Li, R. J. Mural, et al., Science, 2001, 291, 1304–1351.11. P. Germain, B. Staels, C. Dacquet, M. Spedding, and V. Laudet, Pharmacol Rev, 2006, 58, 685–704.12. Nuclear Receptors Nomenclature Committee, Cell, 1999, 97, 161–163.13. L. Jin and Y. Li, Adv Drug Deliv Rev, 2010, 62, 1218–1226.14. D. Metzger, S. Ali, J. M. Bornert, and P. Chambon, J Biol Chem, 1995, 270, 9535–9542.15. S. Khorasanizadeh and F. Rastinejad, Trends Biochem Sci, 2001, 26, 384–390.16. A. Haelens, T. Tanner, S. Denayer, L. Callewaert, and F. Claessens, Cancer Res, 2007, 67, 4514–4523.17. G. A. Peters and S. A. Khan,Molecular Endocrinology, 1999, 13, 286–296.18. D. L. Bain, A. F. Heneghan, K. D. Connaghan Jones, and M. T. Miura, Annu Rev Physiol, 2007, 69,

201–220.19. B. He, S. Bai, A. T. Hnat, R. I. Kalman, J. T. Minges, C. Patterson, and E. M. Wilson, J Biol Chem,

2004, 279, 30643–30653.20. L. Tora, H. Gronemeyer, B. Turcotte, M. P. Gaub, and P. Chambon, Nature, 1988, 333, 185–188.21. M. Leid, P. Kastner, and P. Chambon, Trends Biochem Sci, 1992, 17, 427–433.22. C. A. Sartorius, M. Y. Melville, A. R. Hovland, L. Tung, G. S. Takimoto, and K. B. Horwitz, Mol

Endocrinol, 1994, 8, 1347–1360.23. V. Chandra, P. Huang, Y. Hamuro, S. Raghuram, Y. Wang, T. P. Burris, and F. Rastinejad, Nature,

2008, 456, 350–356.24. V. Kumar, S. Green, G. Stack, M. Berry, J. R. Jin, and P. Chambon, Cell, 1987, 51, 941–951.25. D. L. Bain, M. A. Franden, J. L. McManaman, G. S. Takimoto, and K. B. Horwitz, J Biol Chem, 2001,

276, 23825–23831.26. I. J. McEwan, D. Lavery, K. Fischer, and K. Watt, Nucl Recept Signal, 2007, 5, e001.27. R. Kumar, I. V. Baskakov, G. Srinivasan, D. W. Bolen, J. C. Lee, and E. B. Thompson, J Biol Chem,

1999, 274, 24737–24741.28. D. L. Bain, M. A. Franden, J. L. McManaman, G. S. Takimoto, and K. B. Horwitz, J Biol Chem, 2000,

275, 7313–7320.29. J. R. Wood, V. S. Likhite, M. A. Loven, and A. M. Nardulli,Mol Endocrinol, 2001, 15, 1114–1126.30. J. Brodie and I. J. McEwan, J. Mol Endocrinol, 2005, 34, 603–615.31. K. Dahlman Wright, H. Baumann, I. J. McEwan, T. Almlöf, A. P. Wright, J. A. Gustafsson, and T.

Härd, Proc Natl Acad Sci USA, 1995, 92, 1699–1703.32. J. Reid, S. M. Kelly, K. Watt, N. C. Price, and I. J. McEwan, J Biol Chem, 2002, 277, 20079–20086.33. A. Wärnmark, A. Wikström, A. P. H. Wright, J. Å. Gustafsson, and T. Härd, J Biol Chem, 2001, 276,

45939–45944.34. P. H. Kussie, S. Gorina, V. Marechal, B. Elenbaas, J. Moreau, A. J. Levine, and N. P. Pavletich,

Science, 1996, 274, 948–953.


23

35. M. Uesugi, O. Nyanguile, H. Lu, A. J. Levine, and G. L. Verdine, Science, 1997, 277, 1310–1313.36. M. O. Collins, L. Yu, I. Campuzano, S. G. N. Grant, and J. S. Choudhary,Mol Cell Proteomics, 2008,

7, 1331–1348.37. J. A. Simental, M. Sar, M. V. Lane, F. S. French, and E. M. Wilson, J Biol Chem, 1991, 266, 510–518.38. B. He, J. A. Kemppainen, and E. M. Wilson, J Biol Chem, 2000, 275, 22986–22994.39. T. Ikonen, J. J. Palvimo, and O. A. Jänne, J Biol Chem, 1997, 272, 29821–29828.40. C. A. Berrevoets, P. Doesburg, K. Steketee, J. Trapman, and A. O. Brinkmann, Mol Endocrinol,

1998, 12, 1172–1183.41. T. Hard, E. Kellenbach, R. Boelens, B. A. Maler, K. Dahlman, L. P. Freedman, J. Carlstedt Duke, K.

R. Yamamoto, J. A. Gustafsson, and R. Kaptein, Science, 1990, 249, 157–160.42. B. F. Luisi, W. X. Xu, Z. Otwinowski, L. P. Freedman, K. R. Yamamoto, and P. B. Sigler, Nature,

1991, 352, 497–505.43. J. W. R. Schwabe, L. Chapman, J. T. Finch, and D. Rhodes, Cell, 1993, 75, 567–578.44. R. Kumar and E. B. Thompson, Steroids, 1999, 64, 310–319.45. L. P. Freedman, B. F. Luisi, Z. R. Korszun, R. Basavappa, P. B. Sigler, and K. R. Yamamoto, Nature,

1988, 334, 543–546.46. H. Baumann, K. Paulsen, H. Kovacs, H. Berglund, A. P. H. Wright, J. A. Gustafsson, and T. Haerd,

Biochemistry, 1993, 32, 13463–13471.47. F. Rastinejad, T. Perlmann, R. M. Evans, and P. B. Sigler, Nature, 1995, 375, 203–211.48. S. M. . Holmbeck, M. P. Foster, D. R. Casimiro, D. S. Sem, H. J. Dyson, and P. E. Wright, J Mol Biol,

1998, 281, 271–284.49. M. S. Lee, D. S. Sem, S. A. Kliewer, J. Provencal, R. M. Evans, and P. E. Wright, Eur J Biochem, 1994,

224, 639–650.50. M. D. Gearhart, S. M. A. Holmbeck, R. M. Evans, H. J. Dyson, and P. E. Wright, J Mol Biol, 2003,

327, 819–832.51. W. Zwart, R. de Leeuw, M. Rondaij, J. Neefjes, M. A. Mancini, and R. Michalides, J Cell Sci, 2010,

123, 1253–1261.52. Z. X. Zhou, M. Sar, J. A. Simental, M. V. Lane, and E. M. Wilson, J Biol Chem, 1994, 269, 13115–

13123.53. M. L. Cutress, H. C. Whitaker, I. G. Mills, M. Stewart, and D. E. Neal, J Cell Sci, 2008, 121, 957–968.54. K. Umesono and R. M. Evans, Cell, 1989, 57, 1139–1146.55. A. Haelens, G. Verrijdt, L. Callewaert, V. Christiaens, K. Schauwaers, B. Peeters, W. Rombauts,

and F. Claessens, Biochem J, 2003, 369, 141.56. C. Helsen, S. Kerkhofs, L. Clinckemalie, L. Spans, M. Laurent, S. Boonen, D. Vanderschueren, and

F. Claessens,Mol Cell Endocrinol, 2012, 348, 411–417.57. M. Fu, C. Wang, J. Wang, X. Zhang, T. Sakamaki, Y. G. Yeung, C. Chang, T. Hopp, S. A. W. Fuqua,

E. Jaffray, R. T. Hay, J. J. Palvimo, O. A. Jänne, and R. G. Pestell,Mol Cell Biol, 2002, 22, 3373–3388.58. M. Fu, M. Rao, C. Wang, T. Sakamaki, J. Wang, D. Di Vizio, X. Zhang, C. Albanese, S. Balk, C.

Chang, S. Fan, E. Rosen, J. J. Palvimo, O. A. Jänne, S. Muratoglu, M. L. Avantaggiati, and R. G.Pestell,Mol Cell Biol, 2003, 23, 8563–8575.

59. T. M. Tanner, S. Denayer, B. Geverts, N. Van Tilborgh, S. Kerkhofs, C. Helsen, L. Spans, V.Dubois, A. B. Houtsmuller, F. Claessens, and A. Haelens, Cell Mol Life Sci, 2010, 67, 1919–1927.

60. M. Fu, C. Wang, A. T. Reutens, J. Wang, R. H. Angeletti, L. Siconolfi Baez, V. Ogryzko, M. L.Avantaggiati, and R. G. Pestell, J Biol Chem, 2000, 275, 20853–20860.

61. M. Thomas, N. Dadgar, A. Aphale, J. M. Harrell, R. Kunkel, W. B. Pratt, and A. P. Lieberman, JBiol Chem, 2004, 279, 8389–8395.

62. R. de Leeuw, J. Neefjes, and R. Michalides, Int J Breast Cancer, 2011, 2011, 1–10.

Chapter One

24

63. Y. Cui, M. Zhang, R. Pestell, E. M. Curran, W. V. Welshons, and S. A. W. Fuqua, Cancer Res, 2004,64, 9199–9208.

64. C. C. Williams, A. Basu, A. El Gharbawy, L. M. Carrier, C. L. Smith, and B. G. Rowan, BMCBiochem, 2009, 10, 36.

65. C. Giordano, Y. Cui, I. Barone, S. Ando, M. A. Mancini, V. Berno, and S. A. W. Fuqua, BreastCancer Res Treat, 2010, 119, 71–85.

66. I. Barone, L. Brusco, and S. A. W. Fuqua, Clin Cancer Res, 2010, 16, 2702–2708.67. A. L. Wijayaratne and D. P. McDonnell, J Biol Chem, 2001, 276, 35684–35692.68. C. M. Eakin, M. J. Maccoss, G. L. Finney, and R. E. Klevit, Proc Natl Acad Sci USA, 2007, 104, 5794–

5799.69. G. P. Skliris, Z. J. Nugent, B. G. Rowan, C. R. Penner, P. H. Watson, and L. C. Murphy, Endocr

Relat Cancer, 2010, 17, 589–597.70. W. Zwart, A. Griekspoor, V. Berno, K. Lakeman, K. Jalink, M. Mancini, J. Neefjes, and R.

Michalides, EMBO J, 2007, 26, 3534–3544.71. D. J. Mangelsdorf, C. Thummel, M. Beato, P. Herrlich, G. Schütz, K. Umesono, B. Blumberg, P.

Kastner, M. Mark, P. Chambon, and R. M. Evans, Cell, 1995, 83, 835–839.72. L. Clinckemalie, D. Vanderschueren, S. Boonen, and F. Claessens,Mol Cell Endocrinol, 2012, 358, 1–

8.73. J. D. Chen and R. M. Evans, Nature, 1995, 377, 454–457.74. A. J. Hörlein, A. M. Näär, T. Heinzel, J. Torchia, B. Gloss, R. Kurokawa, A. Ryan, Y. Kamei, M.

Söderström, C. K. Glass, and M. G. Rosenfeld, Nature, 1995, 377, 397–404.75. Y. Brélivet, N. Rochel, and D. Moras,Mol Cell Endocrinol, 2012, 348, 466–473.76. J. C. Nwachukwu and K. W. Nettles, EMBO J, 2012, 31, 251–253.77. I. Orlov, N. Rochel, D. Moras, and B. P. Klaholz, EMBO J., 2012, 31, 291–300.78. P. L. Shaffer, D. P. McDonnell, and D. T. Gewirth, Biochemistry, 2005, 44, 2678–2685.79. K. W. Nettles, J. B. Bruning, G. Gil, J. Nowak, S. K. Sharma, J. B. Hahm, K. Kulp, R. B. Hochberg,

H. Zhou, J. A. Katzenellenbogen, B. S. Katzenellenbogen, Y. Kim, A. Joachimiak, and G. L. Greene,Nat Chem Biol, 2008, 4, 241–247.

80. J. Zhang, M. J. Chalmers, K. R. Stayrook, L. L. Burris, Y. Wang, S. A. Busby, B. D. Pascal, R. D.Garcia Ordonez, J. B. Bruning, M. A. Istrate, D. J. Kojetin, J. A. Dodge, T. P. Burris, and P. R.Griffin, Nat Struct Mol Biol, 2011, 18, 556–563.

81. T. Miyamoto, T. Kakizawa, K. Ichikawa, S. Nishio, T. Takeda, S. Suzuki, A. Kaneko, M. Kumagai,J. Mori, K. Yamashita, T. Sakuma, and K. Hashizume,Mol Cell Endocrinol, 2001, 181, 229–238.

82. W. Bourguet, M. Ruff, P. Chambon, H. Gronemeyer, and D. Moras, Nature, 1995, 375, 377–382.83. J. P. Renaud, N. Rochel, M. Ruff, V. Vivat, P. Chambon, H. Gronemeyer, and D. Moras, Nature,

1995, 378, 681–689.84. R. L. Wagner, J. W. Apriletti, M. E. McGrath, B. L. West, J. D. Baxter, and R. J. Fletterick, Nature,

1995, 378, 690–697.85. B. Durand, M. Saunders, C. Gaudon, B. Roy, R. Losson, and P. Chambon, EMBO J., 1994, 13, 5370–

5382.86. D. Barettino, M. M. Vivanco Ruiz, and H. G. Stunnenberg, EMBO J, 1994, 13, 3039–3049.87. F. Saatcioglu, P. Bartunek, T. Deng, M. Zenke, and M. Karin,Mol Cell Biol, 1993, 13, 3675–3685.88. R. T. Nolte, G. B. Wisely, S. Westin, J. E. Cobb, M. H. Lambert, R. Kurokawa, M. G. Rosenfeld, T.

M. Willson, C. K. Glass, and M. V. Milburn, Nature, 1998, 395, 137–143.89. R. E. Watkins, J. M. Maglich, L. B. Moore, G. B. Wisely, S. M. Noble, P. R. Davis Searles, M. H.

Lambert, S. A. Kliewer, and M. R. Redinbo, Biochemistry, 2003, 42, 1430–1438.90. R. K. Bledsoe, V. G. Montana, T. B. Stanley, C. J. Delves, C. J. Apolito, D. D. McKee, T. G. Consler,

D. J. Parks, E. L. Stewart, T. M. Willson, M. H. Lambert, J. T. Moore, K. H. Pearce, and H. E. Xu,Cell, 2002, 110, 93–105.


25

91. J. P. Renaud and D. Moras, Cell Mol Life Sci, 2000, 57, 1748–1769.92. A. K. Shiau, D. Barstad, J. T. Radek, M. J. Meyers, K. W. Nettles, B. S. Katzenellenbogen, J. A.

Katzenellenbogen, D. A. Agard, and G. L. Greene, Nat Struct Biol, 2002, 9, 359–364.93. J. A. Kallen, J. M. Schlaeppi, F. Bitsch, S. Geisse, M. Geiser, I. Delhon, and B. Fournier, Structure,

2002, 10, 1697–1707.94. B. D. Darimont, R. L. Wagner, J. W. Apriletti, M. R. Stallcup, P. J. Kushner, J. D. Baxter, R. J.

Fletterick, and K. R. Yamamoto, Genes Dev., 1998, 12, 3343–3356.95. E. M. McInerney, D. W. Rose, S. E. Flynn, S. Westin, T. M. Mullen, A. Krones, J. Inostroza, J.

Torchia, R. T. Nolte, N. Assa Munt, M. V. Milburn, C. K. Glass, and M. G. Rosenfeld, Genes Dev,1998, 12, 3357–3368.

96. D. M. Heery, E. Kalkhoven, S. Hoare, and M. G. Parker, Nature, 1997, 387, 733–736.97. C. y Chang, J. D. Norris, H. Grøn, L. A. Paige, P. T. Hamilton, D. J. Kenan, D. Fowlkes, and D. P.

McDonnell,Mol Cell Biol, 1999, 19, 8226–8239.98. K. W. Nettles, J. Sun, J. T. Radek, S. Sheng, A. L. Rodriguez, J. A. Katzenellenbogen, B. S.

Katzenellenbogen, and G. L. Greene,Mol Cell, 2004, 13, 317–327.99. H. A. Ingraham and M. R. Redinbo, Curr Opin Struct Biol, 2005, 15, 708–715.100.A. M. Brzozowski, A. C. W. Pike, Z. Dauter, R. E. Hubbard, T. Bonn, O. Engström, L. Öhman, G.

L. Greene, J. Å. Gustafsson, and M. Carlquist, Nature, 1997, 389, 753–758.101.H. E. Xu, T. B. Stanley, V. G. Montana, M. H. Lambert, B. G. Shearer, J. E. Cobb, D. D. McKee, C.

M. Galardi, K. D. Plunket, R. T. Nolte, D. J. Parks, J. T. Moore, S. A. Kliewer, T. M. Willson, and J.B. Stimmel, Nature, 2002, 415, 813–817.

102.A. K. Shiau, D. Barstad, P. M. Loria, L. Cheng, P. J. Kushner, D. A. Agard, and G. L. Greene, Cell,1998, 95, 927–937.

103.A. Makowski, S. Brzostek, R. N. Cohen, and A. N. Hollenberg,Mol Endocrinol, 2003, 17, 273–286.104. X. Hu and M. A. Lazar, Nature, 1999, 402, 93–96.105. T. Heinzel, R. M. Lavinsky, T. M. Mullen, M. Söderström, C. D. Laherty, J. Torchia, W. M. Yang,

G. Brard, S. D. Ngo, J. R. Davie, E. Seto, R. N. Eisenman, D. W. Rose, C. K. Glass, and M. G.Rosenfeld, Nature, 1997, 387, 43–48.

106.G. Liao, L. Y. Chen, A. Zhang, A. Godavarthy, F. Xia, J. C. Ghosh, H. Li, and J. D. Chen, J BiolChem, 2003, 278, 5052–5061.

107. S. Cheng, S. Brzostek, S. R. Lee, A. N. Hollenberg, and S. P. Balk, Mol Endocrinol, 2002, 16, 1492–1501.

108.D. F. Smith and D. O. Toft,Mol Endocrinol, 1993, 7, 4–11.109. C. L. Smith, Z. Nawaz, and B. W. O’Malley,Mol Endocrinol, 1997, 11, 657–666.110. Y. Shang and M. Brown, Science, 2002, 295, 2465–2468.111. R. T. Gampe, V. G. Montana, M. H. Lambert, G. B. Wisely, M. V. Milburn, and H. E. Xu, Genes

Dev., 2000, 14, 2229–2241.112.D. N. Lavery and I. J. McEwan, Biochemistry, 2008, 47, 3352–3359.113. E. M. Wilson and D. Trauner, ChemInform, 2007, 38, no–no.114. B. He, J. T. Minges, L. W. Lee, and E. M. Wilson, J Biol Chem, 2002, 277, 10226–10235.115. E. Langley, Z. X. Zhou, and E. M. Wilson, J Biol Chem, 1995, 270, 29983–29990.116. E. M. Wilson, in Androgen Action in Prostate Cancer, eds. J. Mohler and D. Tindall, Springer US,

2009, 241–267.117. C. Chang, J. Abdo, T. Hartney, and D. P. McDonnell,Mol Endocrinol, 2005, 19, 2478–2490.118. C. L. Hsu, Y. L. Chen, H. J. Ting, W. J. Lin, Z. Yang, Y. Zhang, L. Wang, C. T. Wu, H. C. Chang, S.

Yeh, S. W. Pimplikar, and C. Chang,Mol Endocrinol, 2005, 19, 350–361.119.D. J. van de Wijngaart, H. J. Dubbink, M. E. van Royen, J. Trapman, and G. Jenster, Mol Cell

Endocrinol, 2012, 352, 57–69.

Chapter One

26

120. E. B. Askew, J. T. Minges, A. T. Hnat, and E. M. Wilson,Mol Cell Endocrinol, 2012, 348, 403–410.121.V. Buzón, L. R. Carbó, S. B. Estruch, R. J. Fletterick, and E. Estébanez Perpiñá,Mol Cell Endocrinol,

2012, 348, 394–402.122. E. Estébanez Perpiñá, L. A. Arnold, A. A. Arnold, P. Nguyen, E. D. Rodrigues, E. Mar, R.

Bateman, P. Pallai, K. M. Shokat, J. D. Baxter, R. K. Guy, P. Webb, and R. J. Fletterick, Proc NatlAcad Sci USA, 2007, 104, 16074–16079.

123. T. W. Moore, C. G. Mayne, and J. A. Katzenellenbogen,Mol Endocrinol, 2010, 24, 683–695.124. J. T. De Leon, A. Iwai, C. Feau, Y. Garcia, H. A. Balsiger, C. L. Storer, R. M. Suro, K. M. Garza, S.

Lee, Y. Sang Kim, Y. Chen, Y. M. Ning, D. L. Riggs, R. J. Fletterick, R. K. Guy, J. B. Trepel, L. M.Neckers, and M. B. Cox, Proc Natl Acad Sci USA, 2011, 108, 11878–11883.

125.D. M. Tanenbaum, Y. Wang, S. P. Williams, and P. B. Sigler, Proc Natl Acad Sci USA, 1998, 95,5998–6003.

126.D. F. Skafar and C. Zhao, Endocrine, 2008, 33, 1–8.127. S. Mosselman, J. Polman, and R. Dijkema, FEBS Letters, 1996, 392, 49–53.128.G. G. J. M. Kuiper, J. G. Lemmen, B. Carlsson, J. C. Corton, S. H. Safe, P. T. van der Saag, B. van

der Burg, and J. Å. Gustafsson, Endocrinology, 1998, 139, 4252–4263.129. E. Treuter, L. Johansson, J. S. Thomsen, A. Wärnmark, J. Leers, M. Pelto Huikko, M. Sjöberg, A. P.

H. Wright, G. Spyrou, and J. Å. Gustafsson, J Biol Chem, 1999, 274, 6667–6677.130.A. Wärnmark, T. Almlöf, J. Leers, J. Å. Gustafsson, and E. Treuter, J Biol Chem, 2001, 276, 23397–

23404.131. R. V. Weatherman and T. S. Scanlan, J Biol Chem, 2001, 276, 3827–3832.132.M. M. Montano, V. Müller, A. Trobaugh, and B. S. Katzenellenbogen,Mol Endocrinol, 1995, 9, 814–

825.133. J. A. Schwartz, L. Zhong, S. Deighton Collins, C. Zhao, and D. F. Skafar, J Biol Chem, 2002, 277,

13202–13209.134. S. P. Williams and P. B. Sigler, Nature, 1998, 393, 392–396.135. B. Farboud, H. Hauksdottir, Y. Wu, and M. L. Privalsky,Mol Cell Biol, 2003, 23, 2844–2858.136. B. Farboud and M. L. Privalsky,Mol Endocrinol, 2004, 18, 2839–2853.137. E. H. Hani, L. Suaud, P. Boutin, J. C. Chèvre, E. Durand, A. Philippi, F. Demenais, N. Vionnet, H.

Furuta, G. Velho, G. I. Bell, B. Laine, and P. Froguel, J Clin Invest, 1998, 101, 521–526.138. F. M. Sladek, M. D. Ruse Jr, L. Nepomuceno, S. M. Huang, and M. R. Stallcup,Mol Cell Biol, 1999,

19, 6509–6522.139.A. A. Bogan, Q. Dallas Yang, M. D. Ruse Jr, Y. Maeda, G. Jiang, L. Nepomuceno, T. S. Scanlan, F.

E. Cohen, and F. M. Sladek, J Mol Biol, 2000, 302, 831–851.140.M. D. Ruse Jr, M. L. Privalsky, and F. M. Sladek,Mol Cell Biol, 2002, 22, 1626–1638.141.M. Anbalagan, B. Huderson, L. Murphy, and B. G. Rowan, Nucl Recept Signal, 2012, 10.142. P. McCall, L. K. Gemmell, R. Mukherjee, J. M. S. Bartlett, and J. Edwards, Br J Cancer, 2008, 98,

1094–1101.143. S. Mukherjee, M. Thomas, N. Dadgar, A. P. Lieberman, and J. A. Iñiguez Lluhí, J Biol Chem, 2009,

284, 21296–21306.144. T. Bawa Khalfe, J. Cheng, Z. Wang, and E. T. H. Yeh, J Biol Chem, 2007, 282, 37341–37349.145. L. C. Murphy, Y. Niu, L. Snell, and P. Watson, Clin Cancer Res, 2004, 10, 5902–5906.146. J. Jiang, N. Sarwar, D. Peston, E. Kulinskaya, S. Shousha, R. C. Coombes, and S. Ali, Clin Cancer

Res, 2007, 13, 5769–5776.147.D. Generali, F. M. Buffa, A. Berruti, M. P. Brizzi, L. Campo, S. Bonardi, A. Bersiga, G. Allevi, M.

Milani, S. Aguggini, M. Papotti, L. Dogliotti, A. Bottini, A. L. Harris, and S. B. Fox, J Clin Oncol,2009, 27, 227–234.


27

148. S. A. Fuqua, C. Wiltschke, Q. X. Zhang, A. Borg, C. G. Castles, W. E. Friedrichs, T. Hopp, S.Hilsenbeck, S. Mohsin, P. O’Connell, and D. C. Allred, Cancer Res, 2000, 60, 4026–4029.

149. C. Wang, M. Fu, R. H. Angeletti, L. Siconolfi Baez, A. T. Reutens, C. Albanese, M. P. Lisanti, B. S.Katzenellenbogen, S. Kato, T. Hopp, S. A. W. Fuqua, G. N. Lopez, P. J. Kushner, and R. G. Pestell,J Biol Chem, 2001, 276, 18375–18383.

150.K. Conway, E. Parrish, S. N. Edmiston, D. Tolbert, C. K. Tse, J. Geradts, C. A. Livasy, H. Singh, B.Newman, and R. C. Millikan, Breast Cancer Res, 2005, 7, R871–880.

151.M. H. Herynk, I. Parra, Y. Cui, A. Beyer, M. F. Wu, S. G. Hilsenbeck, and S. A. W. Fuqua, ClinCancer Res, 2007, 13, 3235–3243.

152. E. Irusen, J. G. Matthews, A. Takahashi, P. J. Barnes, K. F. Chung, and I. M. Adcock, J Allergy ClinImmunol, 2002, 109, 649–657.

153. Z. Szatmáry, M. J. Garabedian, and J. Vilcek, J Biol Chem, 2004, 279, 43708–43715.154.A. L. Miller, M. S. Webb, A. J. Copik, Y. Wang, B. H. Johnson, R. Kumar, and E. B. Thompson,Mol

Endocrinol, 2005, 19, 1569–1583.155.K. Ito, S. Yamamura, S. Essilfie Quaye, B. Cosio, M. Ito, P. J. Barnes, and I. M. Adcock, J Exp Med,

2006, 203, 7–13.156.O. van Beekum, V. Fleskens, and E. Kalkhoven, Obesity, 2009, 17, 213–219.157.G. Pascual, A. L. Fong, S. Ogawa, A. Gamliel, A. C. Li, V. Perissi, D. W. Rose, T. M. Willson, M. G.

Rosenfeld, and C. K. Glass, Nature, 2005, 437, 759–763.158.G. Pascual, A. L. Sullivan, S. Ogawa, A. Gamliel, V. Perissi, M. G. Rosenfeld, and C. K. Glass,

Novartis Found Symp, 2007, 286, 183–196; discussion 196–203.159. C. Jennewein, A. M. Kuhn, M. V. Schmidt, V. Meilladec Jullig, A. von Knethen, F. J. Gonzalez, and

B. Brüne, J Immunol, 2008, 181, 5646–5652.160. J. B. Biggins and J. T. Koh, Curr Opin Chem Biol, 2007, 11, 99–110.161.K. C. Carmony and K. B. Kim,Methods Mol Biol, 2012, 832, 627–638.162.K. M. Sakamoto, K. B. Kim, A. Kumagai, F. Mercurio, C. M. Crews, and R. J. Deshaies, Proc Natl

Acad Sci USA, 2001, 98, 8554–8559.163. J. S. Schneekloth Jr, F. N. Fonseca, M. Koldobskiy, A. Mandal, R. Deshaies, K. Sakamoto, and C. M.

Crews, J Am Chem Soc, 2004, 126, 3748–3754.164.K. Cyrus, M. Wehenkel, E. Y. Choi, H. Lee, H. Swanson, and K. B. Kim, ChemMedChem, 2010, 5,

979–985.165.A. Rodriguez Gonzalez, K. Cyrus, M. Salcius, K. Kim, C. M. Crews, R. J. Deshaies, and K. M.

Sakamoto, Oncogene, 2008, 27, 7201–7211.166.D. Dirnberger, G. Unsin, S. Schlenker, and C. Reichel, ChemBioChem, 2006, 7, 936–942.167. F. B. Perler, E. O. Davis, G. E. Dean, F. S. Gimble, W. E. Jack, N. Neff, C. J. Noren, J. Thorner, and

M. Belfort, Nucleic Acids Res, 1994, 22, 1125–1127.168. T. W. Muir, D. Sondhi, and P. A. Cole, Proc Natl Acad Sci USA, 1998, 95, 6705–6710.169. S. Chong, F. B. Mersha, D. G. Comb, M. E. Scott, D. Landry, L. M. Vence, F. B. Perler, J. Benner, R.

B. Kucera, C. A. Hirvonen, J. J. Pelletier, H. Paulus, and M. Q. Xu, Gene, 1997, 192, 271–281.170. B. C. Kallenberger, J. D. Love, V. K. K. Chatterjee, and J. W. R. Schwabe, Nat Struct Mol Biol, 2003,

10, 136–140.171. S. Möcklinghoff, R. Rose, M. Carraz, A. Visser, C. Ottmann, and L. Brunsveld, ChemBioChem, 2010,

11, 2251–2254.172. S. W. Lockless and T. W. Muir, Proc Natl Acad Sci USA, 2009, 106, 10999–11004.173.G. Palmore and V. Siu, 2011.174. R. Liang, J. Zhou, and J. Liu, Appl Environ Microbiol, 2011, 77, 2488–2495.175.G. Mayer and A. Heckel, Angew Chem Int Ed, 2006, 45, 4900–4921.176. F. G. Cruz, J. T. Koh, and K. H. Link, J Am Chem Soc, 2000, 122, 8777–8778.

Chapter One

28

177. Y. Shi and J. T. Koh, ChemBioChem, 2004, 5, 788–796.178.K. Deuschle, M. Fehr, M. Hilpert, I. Lager, S. Lalonde, L. L. Looger, S. Okumoto, J. Persson, A.

Schmidt, and W. B. Frommer, Cytometry Part A, 2005, 64A, 3–9.179. R. M. Clegg, in FRET and FLIM Techniques, Elsevier, 2008, 1–48.180. P. J. Verveer and Q. S. Hanley, in FRET and FLIM Techniques, Elsevier, 2008, 59–90.181.M. E. van Royen, C. Dinant, P. Farla, J. Trapman, and A. B. Houtsmuller, in The Nuclear Receptor

Superfamily: Methods and Protocols, Springer, New York, 2008, 505, 69–96.182.M. Awais, M. Sato, K. Sasaki, and Y. Umezawa, Anal Chem, 2004, 76, 2181–2186.183.M. Awais, M. Sato, and Y. Umezawa, ChemBioChem, 2007, 8, 737–743.184. L. M. van der Velden, M. V. Golynskiy, I. T. G. W. Bijsmans, S. W. C. van Mil, L. W. J. Klomp, M.

Merkx, and S. F. J. van de Graaf, Hepatology, 2012, in press.185. J. O. Jones and M. I. Diamond, ACS Chem Biol, 2008, 3, 412–418.186.M. E. van Royen, S. M. Cunha, M. C. Brink, K. A. Mattern, A. L. Nigg, H. J. Dubbink, P. J.

Verschure, J. Trapman, and A. B. Houtsmuller, J Cell Biol, 2007, 177, 63–72.187.H. Gronemeyer, J. A. Gustafsson, and V. Laudet, Nat Rev Drug Discov, 2004, 3, 950–964.188. R. F. Witzmann, Steroids: Keys to Life, Van Nostrand Reinhold, New York, 1981.189.V. C. Jordan, Br J Pharmacol, 2006, 147, S269–S276.190. L. H. Sarett, G. E. Arth, R. M. Lukes, R. E. Beyler, G. I. Poos, W. F. Johns, and J. M. Constantin, J

Am Chem Soc, 1952, 74, 4974–4976.191. R. B. Woodward, F. Sondheimer, and D. Taub, J Am Chem Soc, 1951, 73, 4057–4057.192.H. W. Ward, Br Med J, 1973, 1, 13–14.193.W. Sneader, in Drug Discovery A History, John Wiley & Sons Ltd, 2005, 198–199.194. R. C. Heel, R. N. Brogden, T. M. Speight, and G. S. Avery, Drugs, 1978, 16, 1–24.195.M. Clemons, S. Danson, and A. Howell, Cancer Treatment Reviews, 2002, 28, 165–180.196. J. P. Overington, B. Al Lazikani, and A. L. Hopkins, Nat Rev Drug Discov, 2006, 5, 993–996.197.M. C. Via, Nuclear Receptors: The Pipeline Outlook, Cambridge Healthtech Institute, Needham, MA,

USA, 2010.198. F. M. Sladek, Expert Opin Ther Targets, 2003, 7, 679–684.199.A. M. Brzozowski, A. C. W. Pike, Z. Dauter, R. E. Hubbard, T. Bonn, O. Engström, L. Öhman, G.

L. Greene, J. Å. Gustafsson, and M. Carlquist, Nature, 1997, 389, 753–758.200. P. Germain, J. Iyer, C. Zechel, and H. Gronemeyer, Nature, 2002, 415, 187–192.201. E. S. Klein, M. E. Pino, A. T. Johnson, P. J. A. Davies, S. Nagpal, S. M. Thacher, G. Krasinski, and R.

A. S. Chandraratna, J Biol Chem, 1996, 271, 22692–22696.202. T. A. Jackson, J. K. Richer, D. L. Bain, G. S. Takimoto, L. Tung, and K. B. Horwitz, Mol Endocrinol,

1997, 11, 693–705.203. J. L. Oberfield, J. L. Collins, C. P. Holmes, D. M. Goreham, J. P. Cooper, J. E. Cobb, J. M. Lenhard,

E. A. Hull Ryde, C. P. Mohr, S. G. Blanchard, D. J. Parks, L. B. Moore, J. M. Lehmann, K. Plunket,A. B. Miller, M. V. Milburn, S. A. Kliewer, and T. M. Willson, Proc Natl Acad Sci USA, 1999, 96,6102–6106.

204.M. Albers, B. Blume, T. Schlueter, M. B. Wright, I. Kober, C. Kremoser, U. Deuschle, and M. Koegl,J Biol Chem, 2006, 281, 4920–4930.

205. E. M. McInerney and B. S. Katzenellenbogen, J Biol Chem, 1996, 271, 24172–24178.206. E. K. Keeton and M. Brown,Mol Endocrinol, 2005, 19, 1543–1554.207. C. L. Smith and B. W. O’Malley, Endocr Rev, 2004, 25, 45–71.208. Z. Liu, D. Auboeuf, J. Wong, J. D. Chen, S. Y. Tsai, M. J. Tsai, and B. W. O’Malley, Proc Natl Acad

Sci USA, 2002, 99, 7940–7944.209.A. Diez Perez, Arq Bras Endocrinol Metabol, 2006, 50, 720–734.210.M. J. Meegan and D. G. Lloyd, Curr Med Chem, 2003, 10, 181–210.


29

211. L. G. Milroy, L. Nieto, and L. Brunsveld, in Amino Acids, Peptides and Proteins, Royal Society ofChemistry, 2012.

212. B. Vaz, S. Möcklinghoff, and L. Brunsveld, in Nuclear Receptors as Drug Targets, eds. E. Ottow andH. Weinmann, Wiley VCH Verlag GmbH & Co. KGaA, 2008, 25–45.

213. C. A. Lipinski, F. Lombardo, B. W. Dominy, and P. J. Feeney, Adv. Drug Deliv. Rev., 2001, 46, 3–26.214. J. Vagner, H. Qu, and V. J. Hruby, Curr Opin Chem Biol, 2008, 12, 292–296.215.A. A. Parent, J. R. Gunther, and J. A. Katzenellenbogen, J Med Chem, 2008, 51, 6512–6530.216. J. R. Gunther, A. A. Parent, and J. A. Katzenellenbogen, ACS Chem Biol, 2009, 4, 435–440.217. J. Becerril and A. D. Hamilton, Angew Chem Int Ed Engl, 2007, 46, 4471–4473.218.A. B. Williams, P. T. Weiser, R. N. Hanson, J. R. Gunther, and J. A. Katzenellenbogen, Org Lett,

2009, 11, 5370–5373.219.A. Sun, T. W. Moore, J. R. Gunther, M. S. Kim, E. Rhoden, Y. Du, H. Fu, J. P. Snyder, and J. A.

Katzenellenbogen, ChemMedChem, 2011, 6, 654–666.220. J. R. Gunther, Y. Du, E. Rhoden, I. Lewis, B. Revennaugh, T. W. Moore, S. H. Kim, R. Dingledine,

H. Fu, and J. A. Katzenellenbogen, J Biomol Screen, 2009, 14, 181–193.221. L. A. Arnold, E. Estébanez Perpiñá, M. Togashi, N. Jouravel, A. Shelat, A. C. McReynolds, E. Mar,

P. Nguyen, J. D. Baxter, R. J. Fletterick, P. Webb, and R. K. Guy, J Biol Chem, 2005, 280, 43048–43055.

222. J. Y. Hwang, L. A. Arnold, F. Zhu, A. Kosinski, T. J. Mangano, V. Setola, B. L. Roth, and R. K. Guy,J Med Chem, 2009, 52, 3892–3901.

223. P. Sadana, J. Y. Hwang, R. R. Attia, L. A. Arnold, G. Neale, and R. K. Guy, ACS Chem Biol, 2011, 6,1096–1106.

224. B. Dreier and A. Plückthun, in Ribosome Display and Related Technologies : Methods and Protocols,Springer, 2012, vol. 805, 261–286.

225. J. Hanes and A. Plückthun, Proc Natl Acad Sci USA, 1997, 94, 4937–4942.226. C. Zahnd, P. Amstutz, and A. Plückthun, Nat Methods, 2007, 4, 269–279.227. E. Hur, S. J. Pfaff, E. S. Payne, H. Grøn, B. M. Buehrer, and R. J. Fletterick, PLoS Biol, 2004, 2, e274.228.N. B. Mettu, T. B. Stanley, M. A. Dwyer, M. S. Jansen, J. E. Allen, J. M. Hall, and D. P. McDonnell,

Mol Endocrinol, 2007, 21, 2361–2377.229. J. Yang, C. Chang, R. Safi, J. Morgan, D. P. McDonnell, P. J. Fuller, C. D. Clyne, and M. J. Young,

Mol Endocrinol, 2011, 25, 32–43.230. T. Phan, H. D. Nguyen, H. Göksel, S. Möcklinghoff, and L. Brunsveld, Chem Commun (Camb), 2010,

46, 8207–8209.231. B. Vaz, S. Möcklinghoff, S. Folkertsma, S. Lusher, J. de Vlieg, and L. Brunsveld, Chem Commun

(Camb), 2009, 5377–5379.232. J. J. Nestor Jr, Curr Med Chem, 2009, 16, 4399–4418.233.A. M. Leduc, J. O. Trent, J. L. Wittliff, K. S. Bramlett, S. L. Briggs, N. Y. Chirgadze, Y. Wang, T. P.

Burris, and A. F. Spatola, Proc Natl Acad Sci USA, 2003, 100, 11273–11278.234. T. R. Geistlinger and R. K. Guy, J Am Chem Soc, 2003, 125, 6852–6853.235. T. R. Geistlinger, A. C. McReynolds, and R. K. Guy, Chem Biol, 2004, 11, 273–281.236. C. Phillips, L. R. Roberts, M. Schade, R. Bazin, A. Bent, N. L. Davies, R. Moore, A. D. Pannifer, A.

R. Pickford, S. H. Prior, C. M. Read, A. Scott, D. G. Brown, B. Xu, and S. L. Irving, J Am Chem Soc,2011, 133, 9696–9699.

237. C. Féau, L. A. Arnold, A. Kosinski, F. Zhu, M. Connelly, and R. K. Guy, ACS Chem Biol, 2009, 4,834–843.

238.N. A. Lack, P. Axerio Cilies, P. Tavassoli, F. Q. Han, K. H. Chan, C. Feau, E. LeBlanc, E. T. Guns,R. K. Guy, P. S. Rennie, and A. Cherkasov, J Med Chem, 2011, 54, 8563–8573.

239.M. D. Gearhart, L. Dickinson, J. Ehley, C. Melander, P. B. Dervan, P. E. Wright, and J. M.Gottesfeld, Biochemistry, 2005, 44, 4196–4203.

Chapter One

30

240. P. B. Dervan and B. S. Edelson, Curr Opin Struct Biol, 2003, 13, 284–299.241. R. K. Suto, R. S. Edayathumangalam, C. L. White, C. Melander, J. M. Gottesfeld, P. B. Dervan, and

K. Luger, J Mol Biol, 2003, 326, 371–380.242.N. G. Nickols and P. B. Dervan, Proc Natl Acad Sci USA, 2007, 104, 10418–10423.243.D. M. Chenoweth, D. A. Harki, J. W. Phillips, C. Dose, and P. B. Dervan, J Am Chem Soc, 2009, 131,

7182–7188.244. L. H. Wang, X. Y. Yang, X. Zhang, K. Mihalic, Y. X. Fan, W. Xiao, O. M. Z. Howard, E. Appella, A.

T. Maynard, and W. L. Farrar, Nature Medicine, 2004, 10, 40–47.245. L. H. Wang, X. Y. Yang, X. Zhang, P. An, H. J. Kim, J. Huang, R. Clarke, C. K. Osborne, J. K.

Inman, E. Appella, and W. L. Farrar, Cancer Cell, 2006, 10, 487–499.246. J. D. Joseph, B. M. Wittmann, M. A. Dwyer, H. Cui, D. A. Dye, D. P. McDonnell, and J. D. Norris,

Proc Natl Acad Sci USA, 2009, 106, 12178–12183.247. J. D. Norris, J. D. Joseph, A. B. Sherk, D. Juzumiene, P. S. Turnbull, S. W. Rafferty, H. Cui, E.

Anderson, D. Fan, D. A. Dye, X. Deng, D. Kazmin, C. Y. Chang, T. M. Willson, and D. P.McDonnell, Chem Biol, 2009, 16, 452–460.

CHAPTER TWO 4 Investigation into Posttranslational Modifications

of the Estrogen Receptor Hinge Region

Abstract. Posttranslational modifications are known to be important for protein stabilityand activity. Especially in less structured regions they can make a significant impact onstructural changes and thus potentially modulate the activity of certain protein domains. Inthis chapter the structural influence of the acetylation of lysines 266 and 268 in estrogenreceptor (ER ) was investigated. These lysines form part of a linker domain between theDNA and ligand binding domain – namely the ER hinge region – which is known to bemainly unstructured. Molecular dynamics (MD) simulation studies predicted structuralchanges upon lysine acetylation, with each mono acetylated isoform (K*K and KK*)showing different amounts of enhancement in terms of their propensity to form helices.Double lysine acetylation exerted the greatest effect on helix formation. Both CD and NMRstudies of 12 mer peptides synthesized by solid phase peptide synthesis confirmed theseresults. Additionally, CD data suggested the presence of a transient polyproline II (PPII) helixthat is less pronounced for K*K* than for KK. This structural element could be important inboth the exploration of molecular and physiological mechanisms and specific targeting. Theacetyl transferase p300 was capable of acetylating both the unacetylated peptide KK and themono acetylated peptides K*K and KK* without any additional auxiliary protein. Theamount of acetylation observed for both mono acetylated peptides K*K and KK* washowever significantly lower compared with the unacetylated peptide KK accessible fordouble acetylation. The acetylation efficiency of the second lysine is thus impaired by thefirst acetylation event.

Chapter Two

32

2.1. Introduction

The human genome consists of only 20,000 – 25,000 protein coding genes. Thus, theinitially expected or hypothetical number of 2,000,000 genes was far off the mark.1–3 Anexplanation for this is that nature introduces diversity at the protein level of an organism indifferent ways, resulting in two to three orders of magnitude higher complexity compared tothe corresponding genome. Modifications during transcription are one way to introducediversity. Alternative promoters, mRNA editing, or alternate splicing expands the genomefour fold, resulting in a transcriptome of around 100,000 mRNAs.4–6 Further diversification isreached by covalent post translational modifications (PTMs) of proteins.7 This leads to a setof approximately 1,000,000 human proteins which is then closer to the number of expectedgenes. These PTMs can be divided into two classes: peptide backbone cleavage, includingcontrolled proteolysis and protein splicing,8,9 and the enzymatic introduction of chemicalgroups resulting in their covalent binding to residue side chains. From the 20 proteinogenicamino acids, 15 can be diversified with PTMs, with phosphorylation, acetylation,glycosylation, amidation and hydroxylation being the most common reaction types.10

Nuclear receptors (NRs) are transcription factors that play an important role in variouscellular processes.11 PTMs influence both the stability and activity of NRs, and therebyprovide a platform for their in addition to small molecule and protein protein interaction(PPI) based regulatory mechanisms.12 Recently, PTMs could be connected with the onset andprogression of several NR based human diseases.13 For example, a direct link between PTMsand disease could be made for sumoylation, acetylation and phosphorylation of various NRs,including estrogen receptor (ER) and androgen receptor (AR). Hormone refractory prostatecancer (HRPC) has been related to AR phosphorylation (p). Upregulation of the PI3K/Aktpathway results in AR pT308 and pS473 and HRPC development.14 Reduced AR ligandbinding could be correlated to pS215 and pS792, resulting in decreased toxicity of theelongated polyQ stretch in Kennedy’s Disease.15 Enhanced prostate cancer cell growth hasbeen linked to both AR acetylation (K630) and desumoylation (K386 and K520).16,17 In case ofER , tamoxifen therapy in breast cancer treatment was more successful in the case of serinephosphorylation pS118 and pS167,18,19 but resulted in tamoxifen resistance for pS305.20

Sumoylation of certain positions in ER , however, might be useful in the prediction of breastcancer response to endocrine therapy.21 Further connections to PTMs could be made in thecase of insulin resistance in adiposity and inflammatory diseases.22,23 In conclusion, PTMs arean important constituent of NR function and show potential as both disease progressionbiomarkers and predictive patient response markers in terms of NR directed treatments.13

Acetylation, or more specifically N acetyl lysine (Figure 2 1), could be identified for all ofthe different sub classes of NRs, including thyroid hormone receptor like, retinoid Xreceptor like (e.g. HNF 4 ), estrogen receptor like (e.g. ER /AR), and steroidogenic factor

Investigation into Posttranslational Modifications of the Estrogen Receptor Hinge Region

33

like. They are involved in the modulation of various cellular functions, such as stability andactivity of the receptor, sensitivity of hormone response, DNA interaction, and subcellulardistribution.24

Figure 2 1 | Lysine modification by lysine acetyl transferase (KAT) and deacetylase (KDAC)Lysine residues exposed on the surface of a protein can be acetylated by KATs or deacetylated byKDACs, which add or remove acetyl groups at the –amine of the side chain. Acetylationneutralizes the positive charge of the free amine.

The acetylation sequences of NRs are predominantly localized in the flexible hinge region(HR). This region of NRs bridges the DNA binding domain (DBD) with the ligand bindingdomain (LBD) and plays an important role in DNA binding, NR dimerization, and nucleartranslocation (see chapter 3 for more details).25–28 In the case of ER , four lysines (K) areknown to be acetylated: K266, K268, K299, K302 and K303 (Table 2 1, ER ).29,30 The last twowere studied more intensively since they are part of a R/KXKK motif conserved across allNRs members, and because of their proximity to the serine at position 305 (S305), which isimportant for ER activity. Sequence homology of these lysine rich areas adds weight to theidea of a functional importance of NR acetylation (Table 2 1). For example, the directconnection between subcellular localization and acetylation could be shown for HNF 4and AR.31,32

Little is still known about NR structural change and the biological outcome uponintroduction of PTMs and thus there is an urgent need to address this issue with chemicalbiology approaches. Motivated by this current state of affairs, this study is focused on theinfluence of lysine acetylation at positions 266 and 268 (K266 and K268, respectively) on thesecondary structure of the mostly flexible HR (262HR273). Furthermore, the ability of theenzyme p300 to acetylate these small 12 mer peptide sequences of the HR synthesized viasolid phase peptide synthesis (SPPS) – was studied in the absence of any auxiliary peptide orprotein. Kim et al. could identify these positions as targets for p300 acetylation by sequentialdeletions of certain domains of the full length ER. Additionally, mutational studies,specifically mutating K266 or K268 reduced protein acetylation by 30 50%.30 Interestingly,these lysine residues are conserved across multiple species, indicating a degree of relevancefor NR function.30

Chapter Two

34

Table 2 1 | Reported PTM acetylation of NRs

NRs (human) position peptide sequence(R/KXKK consensus)

peptide sequence(other site)

ER 295262

PLMIKRSKKNSLALGRMLKHKRQRDD

AR 626 TLGARKLKKLGNLKGR 487 NLEARKTKKKIKGI

HNF4 111172

QCRYCRLKKCFRAGINGDIRAKKIASIA

SF1 102 DRALKQQKKAQIRARXR 197 GPGAGKRLCAI

TR 106125

QCQLCRFKKCIAVGDSKRVAKRKLI

TR 160 QCQECRFKKCIYVG

LXR78170424

EIRPQKRKKGPAPKSEEQIRLKKLKRQE

LRLQDKKLPPLL

LXR 68161

EEPERKRKKGPAPKSEEQIRKKKIRKQQ

FXR 208 LRKNVKQHADQ

Alignment of NR acetylation motif R/KXKK; bold NR are reported to be acetylated at the R/KXKK motif;bold residues, acetylation recognition motif or acetylation site24,29

2.2. Predicted helical character of model 12 mer peptides

Molecular dynamic (MD) simulations were initially performed to gain information aboutthe native structure of the peptides. For the calculations, an implicit solvent method wasused to simulate the properties on the protein surface getting insights in the structure ofprotein domains upon binding to target proteins. These studies showed a moderate overallhelicity ( and 310 helix) of the 262HR273 of ER .(Figure 2 2B). The introduction of a side chainacetyl group at lysines 266 and/or 268, however, significantly changed the helical character(KK < KK* < K*K < K*K*, Figure 2 2A). Compared to the natural non acetylated sequenceKK, the mono acetylated peptide KK* showed an increase in helical content; especially in theC terminal region of the peptide. K*K also increased the helical content, but, in contrast toKK*, did so in the N terminal region. While the doubly acetylated 12 mer peptide K*K*lacking both positive charged side chain lysine residues revealed the highest degree ofhelical secondary structure during the 20 ns MD simulation (Figure 2 2D).


35

Figure 2 2 | Degree of helical content per residue obtained by MD simulationMolecular dynamics (MD) simulations of the sequence of 262HR273 of ER . Starting from anextended initial conformation, simulations were fully unrestrained and all trajectories weregenerated. Each peptide was simulated for a total of 20 ns 20,000 trajectory snapshots spacedevery 1 ps at 51.9 °C (325 K). The degree of helical content per residue was obtained using theptraj module of AMBER omitting the first 5 ns of each simulation: (A) helical content of naturalnon acetylated KK, 266 mono N acetylated K*K, 268 mono N acetylated K*K, and 266/268doubly N acetylated K*K*. (B) Contribution of different helical motifs ( helix, 310 helix) to totalhelicity. More details in Experimental (1.2.9), MD Simulation and Figure 2 10.

2.3. Solid phase peptide synthesis of 12 mers

Peptides of 12 amino acids lengths and representing the 262HR273 sequence of the HingeRegion in different acetylated forms were prepared on solid phase (Rink Amide MBHAresin), using the fluorenylmethyloxycarbonyl (Fmoc) strategy.36,37 The synthesis wasperformed on a 400 mol scale for the first five amino acids of the peptide sequence. Afterconfirmation of peptide sequence and purity by LC/MS, the resin was then divided into four100 mol batches, to allow for the synthesis of the non , the two mono , and the doublyacetylated peptides. The final yields of the peptides after purification were in the range of 15to 25%. The purity and integrity of the final target 12 mer peptides were confirmed byLC/MS (Figure 2 3). The peptides were subsequently subjected to structural and enzymaticstudies using far UV circular dichroism (CD) (see 2.4), nuclear magnetic resonance (NMR)spectroscopy (see 2.5), and an acetyltransferase assay (see 2.6).

A B

Chapter Two

36

Figure 2 3 | Representative LC/MS analysis of K*K* peptide made by SPPSIon chromatogram (top, TIC), photodiode array chromatogram (bottom, PDA), and thecorresponding mass spectrum of double acetylated K*K* peptide analyzed by liquidchromatography–mass spectrometry (LC/MS). Other peptides in Experimental (1.2.9), SPPS andFigure 2 9.

2.4. Structural analysis by far UV circular dichroism (CD)

Far UV circular dichroism (CD) studies were performed to obtain information about thesecondary structure of the 12 mer peptides and the influence of acetylation in this case. Theamide peptide bonds possess two electronic transition states that are both CD active: the

* transition is primarily responsible for the negative bands at 222 nm, while the *transition results in both a positive band at ~190 nm and the negative band at 208 nm, all ofwhich are characteristic of an helix secondary structure.38

In phosphate buffer (5 mM, pH 7.5) the peptides (50 M) displayed a combination of weakhelical character (negative CD signal at < 200 nm, no signal at 208 nm or 222 nm), and arandom coiled polypeptide. It is known that the addition of 2,2,2 trifluorethanol (TFE)increases the propensity of peptides to form secondary structures.39 Indeed, as a consequenceof the step wise addition of TFE (10% 50%), a shift of the minimum at < 200 nm to higherwavelength (towards 208 nm) and the emergence of a minimum around 222 nm wasobserved (Figure 2 4A). Calculation of the fractional helicity fH showed an increase from5.45% to 16.71%, with the greatest increase witnessed between 10% and 30% (v/v) TFE. Thus,a constant concentration of 30% (v/v) TFE was used to study the different peptides. Theapparent helicity of the non acetylated peptides was found to increase as a function ofacetylation, with the doubly acetylated peptide showing the highest helicity (fH, KK < K*K <KK* < K*K*, Figure 2 4).


37

Figure 2 4 | Influence of TFE and acetylation on helical character of the 12 mer peptides(A) Comparison of circular dichroism (CD) data for peptide KK at 50 M measured in 5 mMphosphate (Pi) buffer (pH 7.5, 20 °C) on varying the amount of additive 2,2,2 trifluorethanol(TFE). (B) Peptides KK, KK*, K*K and K*K* without TFE (condition a) and 30% (v/v) TFE(condition b). Mean residue ellipticity was calculated using equation 1 (Experimental (1.2.9), CD).

The impact of the intermolecular structuring on thermal stability was evaluated bymonitoring the CD effect as a function of temperature. Unexpectedly, these studies revealeda decrease of ellipticity at 222 nm normally correlated with greater helicity – upon anincrease in temperature (Figure 2 5, grey).

Figure 2 5 | Influence of temperature on the helicity of the 12 mer peptidesCircular dichroism (CD) measurements (A) tracking the mean residue ellipticity at 222 nm(MRE200nm) of 200 M of KK ( ) and K*K*( ), respectively, over a temperature range of 5°C90°C; and (B) the full MRE spectrum of at 5°C and 90°C. Mean residue ellipticity (MRE) wascalculated using equation 1 (Experimental (1.2.9), CD)

2.5. Structural analysis using Nuclear Magnetic Resonance (NMR)

The NMR chemical shift perturbation studies were in line with the findings from the CDstudies. Analysis of 1H 13C correlation spectra identified a specific pattern of chemical shiftperturbation for helix stabilization, especially at the two lysine positions. Compared withpeptides with a random coil conformation, sheet peptides exhibit a H (proton of backbonecarbon next to the carbonyl carbon) downfield shift and a C (backbone carbon next to the

A B

1

2

1

2

A B

Chapter Two

38

carbonyl carbon) upfield shift. By contrast, helical conformation results in the oppositeeffect, namely an upfield shift of H and a downfield shift of C .40–42 While not all singleresidues show these characteristic effects of enhanced helicity, there is a significant overalltrend in this direction. Interestingly, this trend is most pronounced at the acetylated lysinepositions (Figure 2 6).

Figure 2 6 | 1H 13C chemical shift perturbation studies(A) Chemical shifts for both the 1H and the 13C nuclei upon acetylation of lysines (K) for eachresidue; (B) an expanded view of the superposition of the 1H 13C heteronuclear single quantumcoherence (HSQC) spectra of the free KK (solid thin contours) and K*K* (dashed contours).Arrows assigned to chemical shifts. HA = 1H ; CA = 13C ; e.g. K5 = lysine (K) at position 5.

2.6. Acetylation by lysine acetyltransferase p300

Protein side chain acetylation by lysine acetyltransferases (KATs) occurs specifically atlysine residues (Ks). Histones are proteins whose activity is strongly regulated via lysineacetylation. The specific KATs for histones are otherwise referred to as histone acetyltransferases (HATs), although HATs can also target proteins other than histones.43 Aversatile method to probe the substrate characteristics of HATs is the usage of enzyme basedassays. In this, the acetyl group transfer is directly coupled to a read out signal, such as anincrease in fluorescence intensity. In the HAT assay used in this present study, the minimumcomponents for peptide acetylation are provided: namely (1) acetyl coenzyme A (acetyl CoA),as a source of transferrable acetyl group and (2) the acetyltransferase p300, which catalyzes thetransfer of an acetyl group from acetyl CoA to form N acetyl lysine.44 As a consequence ofthe transferred acetyl groups the sulfhydryl group of the hydrolyzed CoA SH reacts with7 diethylamino 3 (4’ maleimidophenyl) 4 methylcoumarin (CPM) to yield a fluorescentproduct, which can be detected by excitation at 390 nm and emission at 469 nm (Figure 2 7).45

A B


39

Figure 2 7 | Principle of the HAT assayThe acetyl transferase p300 catalyzes the transfer reaction of the acetyl group from acetyl CoA toa lysine residue, simultaneously releasing CoASH. This reacts with 7 diethylamino 3 (4’maleimidophenyl) 4 methylcoumarin (CPM) to form a highly fluorescent product, which can bedetected by excitation at 390 nm and emission at 469 nm. Thus the amount of transferred acetylgroup is determined by the fluorescent signal.46

The non acetylated peptide resulted in the highest fluorescent signal, indicating that mostacetyl groups were transferred to KK. Interestingly, KK even overpowered the positivecontrol histone H4 peptide, a known substrate for p300.47 As expected, thedouble acetylated K*K* peptide showed no signs of further acetylation, with a fluorescentsignal similar to the control experiment (Figure 2 7). The mono acetylated constructs K*Kand KK* could be acetylated by p300, however, the sum of the transferred acetyl groups toboth K*K and KK* (each with one free lysine residue) did not reach the total amounttransferred to KK (two free lysines). Furthermore, there was a difference in the amount ofmono acetylation of KK* compared to K*K.

Figure 2 8 | Influence of pre acetylated residues on the acetylation capabilities of p300Assay conditions: 50 M Acetyl CoA, 22.2 nM p300, 50 M of peptide; 30 min incubation (RT); Ex375 nm, Em 460 nm. Control, no peptide; H4, histone H4 peptide; KK, non acetylated peptide;K*K, KK*,mono acetylated peptides; K*K*, doubly acetylated peptide.

Chapter Two

40

2.7. Discussion

In a separate study investigating the periodicity of lysines in histones the i, i + 4appearance of lysines in histones was reminiscent of that found in an helix in whichbackbone hydrogen bonds occur between i, i + 4 residues.48 This insight provides someindication of a coherent relationship between lysine acetylation and its contribution tosecondary structure, such as helix formation. Similar to histones, tubulins possessunstructured domains in crystal structures that are known to be important for therecognition of microtubule associated proteins. Interestingly, these domains are decoratedwith diverse posttranslational modifications (PTMs), such as acetylated lysine residues.49

Furthermore, acylation, including acetylation, has a stabilizing function in secondarystructures.50 Another study revealed that the helical content of histone H4 increases uponacetylation at the N position of lysines.51 In this present work, the helical content of KK, a12 residue long peptide based on the sequence of the hinge region (HR) of ER , increasedupon lysine acetylation. Mono acetylated K*K (AcLys266) and KK* (AcLys268) as well asdoubly acetylated K*K* (AcLys266/268) showed a higher propensity to form an helix inMD simulations. Synthesis of the corresponding peptides (SPPS) and subsequent structuralanalysis using CD and NMR confirmed these helix stabilizing findings: upon acetylation ofLys266 and Lys268 both the fractional helicity increased (CD) and the chemical shiftperturbation studies (NMR) showed helix stabilization characteristics in the form of anupfield shift of H and concurrent downfield shift of C nuclei.40–42

In nature, acetylation of lysine residues is performed by acetyltransferases, such as histoneacetyltransferase p300 (p300). Different crystal structures are available showing p300 incomplex with its substrates, such as p53,52 hypoxia inducible factor 1 (HIF 1)53 or nuclearreceptor coactivator 3 (NCoA 3).54 In another study concerning the location of the involvedcomponents, the p300 HAT domain was co crystallized with a bi substrate consisting ofacetyl coenzyme A (acetyl CoA) and lysine (Lys CoA). This crystal structure discovered inproximity to the pocket that facilitates the lysine acetyltransferase reaction an additionalhighly electronegative region.55 That signified that the properties of the amino acids flankingthe target lysine residue play an important role. Sequence alignment of non histone p300substrates verified the hypothesis that positive charged residues are required in proximity(up to three to four residues up and downstream) to the lysine acetylated by p300.55 In thecase of KK, three additional positively charged amino acids (3xArg)are present in the 12 merpeptide next to the two lysines. Thus, from the viewpoint of Lys266, the two arginineresidues, Arg263 and Arg269 are present three residues away, a position that is conservedwith positive amino acids in p300 substrates.56 Furthermore, the other lysine residue(Lys268) also meets this requirement with Arg271 that is capable of forming a salt bridgewith residues in the additional electronegative region of p300. Mutational studies on p300,which removed these negatively charged residues (E1505, D1628, D1625) resulted in a


41

decrease in acetylation ability, underlining the importance of the positively charged flankingresidues in effective peptide acetylation.55 This need for a charge dependent interaction toregulate the modification capabilities of the enzyme could also explain the low degree ofacetylation in the case of K*K and KK*, respectively, compared to KK, as determined by theacetylation assay (HAT). The total amount of transferred acetyl groups from acetyl CoA toboth peptides with one free lysine (K*K, KK*) was less compared with KK (two free lysines).In both cases a positive charge had been removed through the introduction of an acetyllysine (AcLys) during peptide synthesis. The consequent neutralization – in each case tworesidues away from the lysine could negatively influence the substrate binding capabilityand therefore the acetylation efficiency. In general, the positive effect of PTMs on thehelical character of protein domains is also directly connected to the protein stability. In thecase of hypoxia inducible factor 1 alpha (HIF1 ) protein, for instance, protein stabilization isa function of acetylation implemented by p300.57

Kim et al. also studied acetylation of Lys266 and Lys268 of ER using a variety ofbiochemical and cell based approaches. In an acetylation assay consisting of a radiolabeledacetyl CoA they could demonstrate acetylation by p300. However, efficient acetylationrequired SRC2RID/PID as an auxiliary protein, containing both the receptor interaction domain(RID) and p300 interaction domain (PID).58 Furthermore, they identified Lys268 as the majoracetylation site of single acetylation. Although in the HAT assay the distinction between thetwo acetylation sites in the non acetylated KK was not possible, single acetylation of K*K(free Lys268) was not favored over KK* (free Lys266).The temperature induced structural change, observed by CD spectroscopy at high peptide

concentrations – expressed in the form of a decrease in ellipticity at 222 nm (Figure 2 5) isan effect that has been reported for peptides,59 denatured proteins,60 as well as forintrinsically disordered proteins (IDPs) and natively unfolded domains, such as theN terminal domain of p53,61 microtubule associated protein Tau,62 or the extracellulardomain of the nerve growth factor.63 This unusual temperature dependent behaviorcompared to the typical unfolding of globular proteins was initially explained by theformation of helices inducing again partial folding of intrinsically unstructured proteins.64

However, multidimensional NMR spectroscopy studies of different IDPs and peptidesrevealed that transiently formed helices exhibited the highest helical character at lowtemperatures, indicating that the negative CD effect is not a function of helix formation.Whereas, a redistribution of the statistical coil involving a general loss of polyproline II (PPII)helix content is responsible for the change in the CD signal at 222 nm.65 Studies show thatthis left handed helix (9.3 Å/turn, 3 residues/turn) occurs in a frequency of around 5%,resulting in one third of the coil state of proteins.66 Apart from proline rich peptides, otheramino acids including lysines can also adopt PPII helical character due to the electrostaticrepulsion between the side chains.67 In this present work, the 12 mer peptides harbour two

Chapter Two

42

lysine residues and an additional three arginines, resulting in five positive charges atphysiological conditions. Thus a PPII helix conformation is very likely. Interestingly, thetemperature effect on the ellipticity at 222 nm is lower for K*K* compared to KK (Figure 25A). However, the curves were observed to level off at a similar value. This could beexplained by the lower degree of initial PPII helical secondary structure of K*K* resultingfrom a double acetylation of the lysines and thus less electrostatic repulsion. The possibletransient presence of a PPII helix provides further opportunities for potential interaction withother proteins. Kinase activity, for example, can be modulated through the binding of the Srchomology 3 (SH3) domain via PPII helices both intramolecular and in target proteins.68

Furthermore, PTMs, including acetylations,69 are both involved in the regulation of differentcellular processes and are thus connected to various diseases.70 As an example, acetylationwas shown to be important for the regulation of RAS oncogenicity as a stabilizer of certaindomains.71 Counterparts of acetyl lysines are bromodomains (BRDs), which form a specifichydrogen bond with the oxygen of the acetyl carbonyl group. Thus, BRDs have recentlybecome interesting as therapeutic targets blocking their interaction with acetyl lysinecontaining domains.72 Thus, the elucidation of the structure of these domains and how theyare modulated is not only important for the understanding of NR function, but also providesstructural insights that are necessary for successful targeting.


43

2.8. Conclusion

In conclusion, this study of the linker domain between the DBD and the LBD of estrogenreceptor , namely the ER hinge region (HR), and the influence of posttranslationalmodifications (PTMs) has provided structural insights that may help to understand thefunction of what is intrinsically a poorly structured region. Initial MD simulation studiespredicted structural changes upon acetylation of lysines at position 266 and 268. While themono acetylated K*K and KK* each showed different degrees of enhanced propensity toform helices, the doublyacetylated peptides (K*K*) displayed the greatest structuralchange compared to the non acetylated peptide KK.Based on studies of 12 mer peptides synthesized by solid phase peptide synthesis (SPPS),

CD (fractional helicity) and NMR studies (characteristic chemical shifts) confirmed theseresults: namely, a higher helical character as a function of lysine acetylation. The CD dataadditionally hinted at the presence of a transient PPII helix, which was less pronounced forK*K* than for KK. This potentially novel structural insight is not only important in theexploration for molecular and physiological mechanisms but could also be an interestingplatform in specific targeting.Finally, we could also show the acetylation of both mono acetylated (K*K, KK*) and non

acetylated peptide (KK) by the acetyl transferases p300 in the absence of an additionalauxiliary protein. Interestingly, the sum of both acetylations of the mono acetylated peptides(K*K, KK*) were significantly lower compared to the double acetylation of the nonacetylated KK. This means that acetylation of Lys266 and the consequent charge reductiondiminishes the acetylation capacity of the adjacent Lys268 and vice versa. This charge basedacetylation capability of p300 taken together with their dissimilar degree of influence on thedomain secondary structure suggests that these acetylations are important for the sensitiveregulation of cellular processes through a change in the structure of what is otherwise apoorly structured part of the hinge region.

Chapter Two

44

2.9. Experimental

GeneralMD simulations and NMRmeasurements and their interpretations were made by Dr. Lidia Nieto.

If not stated otherwise, chemicals were ordered from SIGMA ALDRICH.

Solid Phase Peptide Synthesis & PurificationThe synthesis of all peptides was performed on solid support using the Fmoc strategy on a Preludepeptide synthesizer (PROTEIN TECHNOLOGIES INC.). If not stated otherwise, 200 mol (1 eq.) of RinkAmide MBHA resin (0.59 mmol/g, NOVABIOCHEM) was swollen for 30 min in 5 mlN methylpyrrolidone (NMP). Washing was performed with NMP or dichloromethane (DCM),deprotection of the amino acids (AA, 0.2 M solution in NMP) was performed with piperidine (Dep.,20% (v/v) solution in NMP), activated with 2 (1H Benzotriazole 1 yl) 1,1,3,3 tetramethyluroniumhexafluorophosphate (Act., HBTU, 0.4 M solution in NMP) and alkalized with N,NDiisopropylethylamine (Base, DIPEA, 1.6 M in NMP). The amino acids were double coupled with asubsequent capping step (Cap., 20% (v/v) pyridine, 20% (v/v) acetic anhydride in NMP): 6 ml NMP,4 ml Dep. (5 min), 2x[6 ml NMP, 4 ml AA, 2 ml Act., 2 ml Base, 20 min], 6 ml NMP, 3 ml Cap. (5 min),3 ml NMP. After finishing the desired peptide sequence, the resin was washed 3 times alternately withdiethylether (DEE) and dichloromethane (DCM) and dried under vacuum for 10 min. The side chainprotection group and the cleavage of the peptide sequence from the resin were simultaneouslyachieved by the usage of mixture of trifluoroacetic acid, triisopropylsilane and water (TFA/TIS/H2O,95/2.5/2.5) for 3 h. The free peptide was precipitated by adding it drop wise to ice cold DEE. A finalcentrifugation step (2.000 rpm, 10 min) with subsequent washing (ice cold DEE), uptake in water andlyophilization results in the crude peptide. Purification was accomplished by reverse phase HPLC ona Alltima™ HP C18 column (125x20 mm, Alltech). Water (0.1% TFA) was used as a polar phase, addingdifferent amounts of an apolar acetonitrile (ACN, 0.1% TFA) phase. A linear gradient (20 ml/min) wasoptimized for each peptide to a 5% range, for instance 35 40% ACN. The purity of the peptide wasdeterminate at an analytical LC MS using GraceSmart RP18 (50x2.1mm, Grace, 3u, 120A) columnFinally, the peptides were lyophilized and stored at 80 °C.


45

Figure 2 9 | LC/MS analysis of peptides made by SPPSIon chromatogram (top, TIC), photodiode array (bottom, PDA), and corresponding massspectrum of unacetylated KK (A), mono acetylated K*K (B) and KK* peptide (C), and doubleacetylated K*K* (D) peptide analyzed by liquid chromatography–mass spectrometry (LC/MS).

Circular Dichroism spectroscopy73,74

Far UV Circular Dichroism (CD) spectroscopy measurements were performed under a constantnitrogen flow at 20 °C using 50 M of the peptide in the presence of 30% (v/v) trifluoroethanol (TFE),if not stated otherwise. As a baseline a peptide free buffer (30% TFE) was measured under identicalconditions. The spectrum was recorded from 250 nm to 185 nm using a JASCO 815 spectrometer.Quartz cuvettes with path lengths of 1 mm or 0.2 mm (HELLMA ANALYTICS) were employed with adata pitch of 0.5 in a continuous mode, a scan speed of 20 nm/min with a response time of 2 s and abandwidth of 0.5 nm. The graphs are representing an average of five scans. The observed ellipticity(degrees, in [mdeg]) was converted to mean residue ellipticity (in [deg cm2/dmol]) usingequation 1.

Where is the molecular mass of the peptide (in [Da]), is the number of amino acids in the peptide,is the path length (in [cm]), and is the peptide concentration (in [g/ml]). The fractional helicity ( )

was calculated using equation 2.

Chapter Two

46

Where is the measured mean residue ellipticity at 222 nm, is temperature in °C, andis the number of amide bonds in the peptide.75

Molecular Dynamics SimulationAll MD simulations were carried out using the AMBER suite of programs, and the ff03 force field.34 Animplicit solvent was used via the General Born solvation method (IGB 5), as implemented in AMBER,providing an approximation of the electrostatic free energy when a charge distribution is created in acavity of low dielectric (protein, = 1) embedded in a uniform dielectric medium (water, w = 80).33–35

Starting with an extended initial conformation, built by the LEaP module of AMBER for each peptide,all MD simulations were fully unrestrained and all trajectories were generated using the sanderprogram in the AMBER 9 package. Each peptide was simulated for a total of 20 ns at 51.9 °C, followinga similar approach as previously reported to predict the conformation of a miniprotein.76 Analysis wasperformed on the 20 ns using 20,000 trajectory snapshots spaced every 1 ps. A snapshot with thelowest potential energy across the simulation was chosen as representative structure for each peptide.The secondary structure of each residue as a function of time was subsequently analyzed utilizing theSTRIDE secondary structure assignment algorithms as implemented in VMD.77,78

Figure 2 10 | Degree of helical content per residue based on MD simulationStarting from an extended initial conformation, MD simulations were fully unrestrained and alltrajectories were generated. Each peptide was simulated for a total of 20 ns 20,000 trajectorysnapshots spaced every 1 ps at 51.9 °C (325 K). Grayscale is used to represent the propensity offorming secondary structures: white, unordered; light grey, helix; grey, turn; black, 310 helix.Dashed line indicate the start/end of a helix. A snapshot of the lowest potential energy across thesimulation was chosen as representative structure for each peptide: (A) 266/268 double Nacetylated ER HR262 273 (K*K*). (B) The degree of helical content per residue (for K*K*) wasobtained using the ptrajmodule of AMBER omitting the first 5 ns of each simulation.

A B


47

Nuclear Magnetic Resonance spectroscopySamples of peptides for NMR experiments were dissolved in 5mM PBS buffer (pH 7.4) with 30% TFE(v/v) (trifluoroethanol d3, Cambridge Isotope Laboratories) to give a final peptide concentration of1 mM. Spectra were recorded at 283 K on a Bruker 600 spectrometer (1H frequency of 600 MHz) byusing a 5 mm triple resonance Z gradient probe and processed by using Topspin software. One andtwo dimensional spectra were acquired by using standard pulse sequences and WATERGATE basedsolvent suppression sequences. A standard suite of homonuclear two dimensional experiments totalcorrelation spectroscopy (TOCSY) and nuclear overhauser effect spectroscopy (NOESY) with mixingtimes of 70 ms and 150 ms, respectively, were acquired for chemical shift assignments. To completethe heteronuclear carbon chemical shift assignments, natural abundance 1H 13C heteronuclear singlequantum coherence (HSQC) spectra were acquired. The data were processed in Topspin 2.1 fromBruker Biospin and analyzed using CARA1.5.

Acetylation Assay45,46

The Histone AcetylTransferase (HAT) assay kit (ACTIVE MOTIF) was used, including acetyl CoA, as asource for the acetyl group and histone acetyltransferase p300 that catalyze enzymatically the transferof the acetyl group from acetyl CoA to form N acetyl lysine. The amount of transferred acetylgroups is directly correlated to the fluorescence signal, since the sulfhydryl groups of the acetyl CoA(CoA SH) can react with a provided substrate to a fluorescent product.The assay was performed in 96 well plates with a final volume of 50 l containing 50 M Acetyl CoA,22.2 nM p300 and 50 M of peptide. After 30 min of incubation at room temperature 50 l of stopsolution and 100 l developing solution was added and incubated for 15 min in the dark.Subsequently fluorescence was measured with 375 nm excitation and 460 nm emission.

2.10. References1. E. S. Lander, L. M. Linton, B. Birren, C. Nusbaum, M. C. Zody, et al., Nature, 2001, 409, 860–921.2. J. C. Venter, M. D. Adams, E. W. Myers, P. W. Li, R. J. Mural, et al., Science, 2001, 291, 1304–1351.3. I. H. G. S. Consortium, Nature, 2004, 431, 931–945.4. T. A. Ayoubi and W. J. Van De Ven, FASEB J, 1996, 10, 453–460.5. J. Ashkenas, Am J Hum Genet, 1997, 60, 278–283.6. T. Maniatis and B. Tasic, Nature, 2002, 418, 236–243.7. C. Walsh, Posttranslational Modification of Proteins: Expanding Nature’s Inventory, Roberts and Co.

Publishers, 1st Edition., 2005.8. C. J. Noren, J. Wang, and F. B. Perler, Angew Chem, 2000, 112, 458–476.9. M. E. Fortini, Nat Rev Mol Cell Bio, 2002, 3, 673–684.10. G. A. Khoury, R. C. Baliban, and C. A. Floudas, Scientific Reports, 2011, 1.11. V. Laudet and H. Gronemeyer, Eds., The Nuclear Receptor FactsBook, Academic Press, 1st edn.,

2001.12. J. T. Moore, J. L. Collins, and K. H. Pearce, ChemMedChem, 2006, 1, 504–523.13. M. Anbalagan, B. Huderson, L. Murphy, and B. G. Rowan, Nucl Recept Signal, 2012, 10.14. Y. Liao, R. Grobholz, U. Abel, L. Trojan, M. S. Michel, P. Angel, and D. Mayer, Int J Cancer, 2003,

107, 676–680.15. I. Palazzolo, B. G. Burnett, J. E. Young, P. L. Brenne, A. R. La Spada, K. H. Fischbeck, B. W.

Howell, and M. Pennuto, HumMol Genet, 2007, 16, 1593–1603.16. M. Fu, M. Rao, C. Wang, T. Sakamaki, J. Wang, D. Di Vizio, X. Zhang, C. Albanese, S. Balk, C.

Chang, S. Fan, E. Rosen, J. J. Palvimo, O. A. Jänne, S. Muratoglu, M. L. Avantaggiati, and R. G.Pestell,Mol Cell Biol, 2003, 23, 8563–8575.

Chapter Two

48

17. H. Poukka, U. Karvonen, O. A. Janne, and J. J. Palvimo, Proc Natl Acad Sci USA, 2000, 97, 14145–14150.

18. L. C. Murphy, Y. Niu, L. Snell, and P. Watson, Clin Cancer Res, 2004, 10, 5902–5906.19. H. Yamashita, M. Nishio, S. Kobayashi, Y. Ando, H. Sugiura, Z. Zhang, M. Hamaguchi, K. Mita, Y.

Fujii, and H. Iwase, Breast Cancer Res, 2005, 7, R753–764.20. C. Holm, M. Kok, R. Michalides, R. Fles, R. H. T. Koornstra, J. Wesseling, M. Hauptmann, J.

Neefjes, J. L. Peterse, O. Stål, G. Landberg, and S. C. Linn, J Pathol, 2009, 217, 372–379.21. M. V. Karamouzis, P. A. Konstantinopoulos, F. A. Badra, and A. G. Papavassiliou, Breast Cancer

Res Treat, 2008, 107, 195–210.22. J. M. Abbinante Nissen, L. G. Simpson, and G. D. Leikauf, Am J Physiol, 1995, 268, L601–606.23. O. van Beekum, V. Fleskens, and E. Kalkhoven, Obesity, 2009, 17, 213–219.24. C. Wang, L. Tian, V. M. Popov, and R. G. Pestell, J Steroid Biochem Mol Biol, 2011, 123, 91–100.25. Z. X. Zhou, M. Sar, J. A. Simental, M. V. Lane, and E. M. Wilson, J Biol Chem, 1994, 269, 13115–

13123.26. S. Khorasanizadeh and F. Rastinejad, Trends Biochem Sci, 2001, 26, 384–390.27. G. Verrijdt, A. Haelens, and F. Claessens,Mol Genet Metab, 2003, 78, 175–185.28. A. Haelens, T. Tanner, S. Denayer, L. Callewaert, and F. Claessens, Cancer Res, 2007, 67, 4514–4523.29. C. Wang, M. Fu, R. H. Angeletti, L. Siconolfi Baez, A. T. Reutens, C. Albanese, M. P. Lisanti, B. S.

Katzenellenbogen, S. Kato, T. Hopp, S. A. W. Fuqua, G. N. Lopez, P. J. Kushner, and R. G. Pestell,J Biol Chem, 2001, 276, 18375–18383.

30. M. Y. Kim, E. M. Woo, Y. T. E. Chong, D. R. Homenko, and W. L. Kraus, Mol Endocrinol, 2006, 20,1479–1493.

31. E. Soutoglou, N. Katrakili, and I. Talianidis,Mol Cell, 2000, 5, 745–751.32. M. Thomas, N. Dadgar, A. Aphale, J. M. Harrell, R. Kunkel, W. B. Pratt, and A. P. Lieberman, J

Biol Chem, 2004, 279, 8389–8395.33. B. K. Ho and K. A. Dill, PLoS Comput Biol, 2006, 2, e27.34. Y. Duan, C. Wu, S. Chowdhury, M. C. Lee, G. Xiong, W. Zhang, R. Yang, P. Cieplak, R. Luo, T.

Lee, J. Caldwell, J. Wang, and P. Kollman, J Comput Chem, 2003, 24, 1999–2012.35. A. Onufriev, D. Bashford, and D. A. Case, Proteins, 2004, 55, 383–394.36. E. Atherton, C. Bury, R. C. Sheppard, and B. J. Williams, Tetrahedron Lett, 1979, 20, 3041–3042.37. W. Chan and P. White, Fmoc Solid Phase Peptide Synthesis: A Practical Approach, Oxford University

Press, U.S.A., 2000.38. N. Sreerama and R. Woody, Eds., in Circular Dichroism: Principles and Applications, Wiley VCH,

2nd edn., 2000, 601–620.39. J. K. Myers, C. N. Pace, and J. M. Scholtz, Protein Sci, 1998, 7, 383–388.40. S. Spera and A. Bax, J. Am. Chem. Soc., 1991, 113, 5490–5492.41. D. S. Wishart, B. D. Sykes, and F. M. Richards, Biochemistry, 1992, 31, 1647–1651.42. F. Avbelj, D. Kocjan, and R. L. Baldwin, PNAS, 2004, 101, 17394–17397.43. M. W. Swank and J. D. Sweatt, J Neurosci, 2001, 21, 3383–3391.44. M. A. Glozak, N. Sengupta, X. Zhang, and E. Seto, Gene, 2005, 363, 15–23.45. T. O. Sippel, J Histochem Cytochem, 1981, 29, 314–316.46. C. C. Chung, K. Ohwaki, J. E. Schneeweis, E. Stec, J. P. Varnerin, P. N. Goudreau, A. Chang, J.

Cassaday, L. Yang, T. Yamakawa, O. Kornienko, P. Hodder, J. Inglese, M. Ferrer, B. Strulovici, J.Kusunoki, M. R. Tota, and T. Takagi, Assay Drug Dev Technol, 2008, 6, 361–374.

47. M. Shogren Knaak, H. Ishii, J. M. Sun, M. J. Pazin, J. R. Davie, and C. L. Peterson, Science, 2006,311, 844–847.

48. B. D. Strahl and C. D. Allis, Nature, 2000, 403, 41–45.49. E. Nogales, M. Whittaker, R. A. Milligan, and K. H. Downing, Cell, 1999, 96, 79–88.50. B. C. Poschner and D. Langosch, Protein Sci, 2009, 18, 1801–1805.51. X. Wang, S. C. Moore, M. Laszckzak, and J. Ausió, J Biol Chem, 2000, 275, 35013–35020.


49

52. H. Feng, L. M. Miller Jenkins, S. R. Durell, R. Hayashi, S. J. Mazur, S. Cherry, J. E. Tropea, M.Miller, A. Wlodawer, E. Appella, and Y. Bai, Structure, 2009, 17, 202–210.

53. S. J. Freedman, Z. Y. J. Sun, F. Poy, A. L. Kung, D. M. Livingston, G. Wagner, and M. J. Eck, ProcNatl Acad Sci USA, 2002, 99, 5367–5372.

54. S. J. Demarest, M. Martinez Yamout, J. Chung, H. Chen, W. Xu, H. J. Dyson, R. M. Evans, and P. E.Wright, Nature, 2002, 415, 549–553.

55. X. Liu, L. Wang, K. Zhao, P. R. Thompson, Y. Hwang, R. Marmorstein, and P. A. Cole, Nature,2008, 451, 846–850.

56. P. R. Thompson, H. Kurooka, Y. Nakatani, and P. A. Cole, J Biol Chem, 2001, 276, 33721–33729.57. H. Geng, Q. Liu, C. Xue, L. David, T. M. Beer, G. V. Thomas, M. S. Dai, and D. Z. Qian, J Biol

Chem, 2012.58. W. L. Kraus, E. T. Manning, and J. T. Kadonaga,Mol Cell Biol, 1999, 19, 8123–8135.59. F. Eker, K. Griebenow, and R. Schweitzer Stenner, J Am Chem Soc, 2003, 125, 8178–8185.60. D. Nettels, S. Müller Späth, F. Küster, H. Hofmann, D. Haenni, S. Rüegger, L. Reymond, A.

Hoffmann, J. Kubelka, B. Heinz, K. Gast, R. B. Best, and B. Schuler, Proc Natl Acad Sci USA, 2009,106, 20740–20745.

61. R. Dawson, L. Müller, A. Dehner, C. Klein, H. Kessler, and J. Buchner, J Mol Biol, 2003, 332, 1131–1141.

62. S. Jeganathan, M. von Bergen, E. M. Mandelkow, and E. Mandelkow, Biochemistry, 2008, 47,10526–10539.

63. D. E. Timm, P. Vissavajjhala, A. H. Ross, and K. E. Neet, Protein Sci, 1992, 1, 1023–1031.64. V. N. Uversky, Eur J Biochem, 2002, 269, 2–12.65. M. Kjaergaard, A. B. Nørholm, R. Hendus Altenburger, S. F. Pedersen, F. M. Poulsen, and B. B.

Kragelund, Protein Sci, 2010, 19, 1555–1564.66. Y. Mansiaux, A. P. Joseph, J. C. Gelly, and A. G. de Brevern, PLoS ONE, 2011, 6, e18401.67. A. L. Rucker and T. P. Creamer, Protein Sci, 2002, 11, 980–985.68. W. Gmeiner, J. Xu, D. Horita, T. Smithgall, J. Engen, D. Smith, and R. Byrd, Cell Biochem Biophys,

2001, 35, 115–126.69. K. L. Guan and Y. Xiong, Trends Biochem Sci, 2011, 36, 108–116.70. A. Iyer, D. P. Fairlie, and L. Brown, Immunol Cell Biol, 2012, 90, 39–46.71. M. H. Yang, S. Nickerson, E. T. Kim, C. Liot, G. Laurent, R. Spang, M. R. Philips, Y. Shan, D. E.

Shaw, D. Bar Sagi, M. C. Haigis, and K. M. Haigis, Proc Natl Acad Sci USA, 2012, 109, 10843–10848.72. S. Muller, P. Filippakopoulos, and S. Knapp, Expert Rev Mol Med, 2011, 13.73. S. M. Kelly, T. J. Jess, and N. C. Price, Biochim Biophys Acta, 2005, 1751, 119–139.74. N. J. Greenfield, Nat Protoc, 2006, 1, 2527–2535.75. S. M. Gregory, E. Harada, B. Liang, S. E. Delos, J. M. White, and L. K. Tamm, Proc Natl Acad Sci

USA, 2011, 108, 11211–11216.76. C. Simmerling, B. Strockbine, and A. E. Roitberg, J Am Chem Soc, 2002, 124, 11258–11259.77. D. Frishman and P. Argos, Proteins, 1995, 23, 566–579.78. W. Humphrey, A. Dalke, and K. Schulten, J Mol Graph, 1996, 14, 33–38, 27–28.

CHAPTER THREE 4 Studying the Estrogen Receptor Hinge Region

Using Expressed Protein Ligation

Abstract. Protein engineering has long been limited to mutational studies focusing on the20 proteinogenic amino acids. The introduction of protein (semi )synthesis techniques hasenabled access to a much broader set of proteins with chemical probes and modificationssuch as posttranslational modifications (PTMs). In this present study, expressed proteinligation (EPL) was used to investigate the hinge region (HR) of the estrogen receptor (ER )and its PTMs in the context of their impact on the ligand binding domain (LBD). Using theamino terminal methionine excision approach, it was possible to generate recombinant ERLBD with an N’ terminal cysteine, required for the EPL reaction. Thioester peptides ascounterparts to the N’ terminal cysteine in the EPL were synthesized by solid phasepeptide synthesis using the Dawson method, which uses a fluorenylmethyloxy carbonylbased coupling strategy in combination with a diaminobenzoyl linker attached to a Rinkamide resin. While ligation of the posttranslationally modified HR peptides to the modelprotein C YFP could be successfully demonstrated, the transfer of this method to ER wasnot successful. Protein instability under ligation conditions proved to be the major issue inthis case. Studies focusing on the extension of the ER LBD in the direction of the HRrevealed that protein expression allowed the generation of ER LBD constructs including theHR up to position 251. Binding studies identified the HR sequence before position 271 tosignificantly decrease the ability of ER to bind coactivator peptides, thus hinting at theexistence of two regions of the HR, which differ in terms of their structure and function.

CHAPTER THREE

52

3.1. Introduction

Protein engineering plays a crucial role in biochemistry, biomedicine and biotechnology.1

Classical mutational studies have one essential limitation, however: namely, the introductionof solely 20 genetically encoded amino acids. This means that natural ribosomal synthesis isincapable of introducing unnatural amino acids, biophysical probes, or posttranslationalmodifications (PTMs). However, in the last decades different techniques have beendeveloped for the specific modification of proteins, including unnatural amino acidmutagenesis and native chemical ligation (NCL). These are two powerful techniques tocircumvent the limitations of natural protein synthesis and have multiple applications.2,3

While modifications are introduced at the stage of translation in the case of the mutagenesismethod, NCL combines two fragments of peptides or proteins in a chemoselective manner.4

This requires a carboxy(C’) terminal thioester group and an amino(N’) terminal cysteine(N’ Cys), resulting in a native peptide bond bearing a cysteine residue at the junction (Figure3 1). Main advantages of this chemical peptide synthesis technique are that the peptide canbe fully unprotected and that the reaction can be performed under neutral conditions. In the1950s the synthesis of small peptides had already been performed using thioester chemistryand thus Wieland et al. can be viewed as pioneers in the field of NCL.5 The introduction ofthis specific ligation reaction resulted in various chemical syntheses of chemically modifiedpeptides of medium size. Standard peptide synthesis on solid phase (SPPS) is typicallylimited to peptide sequences of no greater than 60 residues. NCL, however, can be used tocouple large peptide fragments to generate even larger proteins.6 During this type of proteinsemi synthesis, also named expressed protein ligation (EPL), a recombinantly expressedprotein containing an thioester is ligated via NCL to a chemically synthesized peptidecontaining an N’ Cys or vice versa (with thioester on the chemically synthesized peptide andN’ Cys on the recombinant expressed protein)4,7 A wide range of chemical modificationshave been introduced using EPL on different classes of proteins, including antibodies, ionchannels, polymerases, phosphatases, kinases, signaling proteins and transcription factorslike nuclear receptors (NRs).8,9

Studying the Estrogen Receptor Hinge Region Using Expressed Protein Ligation

53

The following six areas of application have in particular benefited from this new technique:

site specific solid support immobilizationo e.g. covalent attachment of proteins onto glass surfaces10

polypeptide backbone cyclizationo e.g. biosynthesis of circular peptides and

proteins using intramolecular EPL reaction in vitro and in vivo11

isotope edited spectroscopyo e.g. decreasing spectral complexity via

selective residue labeling with NMR active nuclei12

incorporation of non natural amino acidso e.g. such as the semi synthesis of an ion channel

using D amino acids to modify selectivity13

incorporation of optical probeso e.g. synthesis of doubly fluorescently labeled protein

to study stability in living cells by microscopic imaging14

introduction of PTMs for structural and biological studieso e.g. site specific protein prenylation,15 acetylation16 or phosphorylation17

The preparation of the starting products for the EPL reaction is an essential aspect of thisapproach for which a number of different methods are currently available. The generation ofa recombinant protein with an N’ terminal cysteine residue (N’ Cys) can be achieved usingdifferent techniques. The introduction of cleavage sites of exogenous proteases (Factor Xa,TEV or thrombin) directly prior to a cysteine enables purification via an N’ terminal tag andsubsequent cleavage with the exogenous proteases, resulting in an N’ Cys.18–20 Anotheroption is to generate an N’ Cys by using endogenous leader peptidases in the periplasmicspace to which an N’ terminal pelB leader sequence is directing.21–23 Furthermore, theapproach of N’ terminal methionine excision (NME) can be used. In this case the endogenousmethionyl aminopeptidases removes the methionine during expression in E. coli, generatingan N’ Cys.24,25 Finally, protein splicing and inteins can be used to produce N’ Cys. Severalinteins such as Mxe GyrA, Mth RIR1 and Ssp DnaB have been modified so that splicing canbe accomplished in a pH and temperature dependent fashion to provide an N’ Cys.26–28

The counterpart to the N’ Cys bearing protein in EPL, a thioester featuring peptide, can besynthesized by solid phase peptide synthesis (SPPS) using a wide range of approaches,including mercaptopropionamide , sulfonamide , backbone amide , S protectedoxazolidinone , aryl hydrazine and mercaptocarboxyethylester linkers,7,29–33 as well as Lewis

CHAPTER THREE

54

acid activated cleavage and the introduction of an N acyl benzimidazolinone (Nbz) leavinggroup.34,35

In this study EPL was used to investigate the hinge region (HR) of the estrogen receptor(ER ) and its PTMs – in particular acetylation on the lysine residues at position 266 and 268(Lys266, Lys268) in the context of the entire ligand binding domain (Figure 3 1). The N’ Cysof the recombinant construct was generated using the NME approach, while the peptideswere synthesized using fluorenylmethyloxycarbonyl(Fmoc) based SPPS on the Dawson resinwith a diaminobenzoyl (Dbz) linker.35 Activation of the peptide led to the correspondingN acyl benzimidazolinone (Nbz) leaving group, which was then converted to a reactivethioester during NCL. But aforetime, the ligand binding domain (LBD) of ER was extendedprogressively in the N’ terminal direction towards the HR to gain information about itsinfluence on protein stability. Furthermore, the variant expressing the entire HR can be seenas the recombinant counterpart to the semi synthesized expressed LBD ligated to thechemically synthesized HR bearing posttranslational modifications.

Figure 3 1 | Semi synthesis of ER LBD and PTM modified HR by the EPL reactionExpressed protein ligation (EPL) of the recombinantly expressed LBD of ER (283ER LBD553) topeptides (262HR281) synthesized by solid phase peptide synthesis (SPPS), including PTMs(acetylation, Ac). N’ terminal methionine is cleaved off by endogenous methionylaminopeptidases (NME) during expression in E. coli. Release of N acyl benzimidazolinone (Nbz)leads to the C’ terminal peptide thioester. Transthioesterification and subsequent S,N acyl shiftproduces a native peptide bond between the chemically synthesized peptide and therecombinantly expressed protein.

Chemical Synthesis Recombinant Expression


55

3.2. Results and discussion

3.2.1.Generating the recombinant protein for EPL

The EPL at the N’ terminus of the LBD of ER (around position 300) requires anN’ terminal cysteine (N’ Cys), one of the least abundant residues in proteins.36 The cysteineclosest to the LBD, however, is located at position 381 and is thus impractical for thispurpose. Site specific mutation of a suitably located amino acid into a cysteine preferablywith similar properties is an alternative option. Serine is ideally suited for this purposesince it differs from cysteine only in terms of one atom (oxygen versus sulfur) and isbiosynthetically related. Serine occurs at a higher frequency than cysteine and in the case ofthe N’ terminus of ER LBD it can be found at positions 282, 294, 301, 305, and 309. For thiswork the introduction of PTMs at position 266 and 268 serine at position 282 (Ser282) ismost convenient. Site directed mutagenesis resulted in the following construct: ER281(S282C)HR LBD553 FXa His6. This plasmid was transformed into E. coli BL21 expressioncells and protein was expressed for 20 h at 15 °C. During expression endogenous methionylaminopeptidases was envisaged to remove Met281, resulting in a free N’ Cys (C ER /S282C,Figure 3 1, right). Purification was performed by means of the C’ terminal His6 tag via NiNTA affinity chromatography. SDS analysis of the protein showed a high purity afterpurification (Figure 3 2), however, the final yield of pure protein was low (0.5 mg/L).

Figure 3 2 | SDS PAGE of expression and purification of ER LBD construct for EPLSodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS PAGE) of ER 281(S282C)HRLBD553 FXa His6 expression in E. coli (TB medium, 15 °C, 15 h, 10 M estradiol, E2) andpurification on a nitrilotriacetic acid based nickel (Ni NTA) column leads to pure C ER /S282C inthe 250 mM imidazole fraction. Other fractions: 100 mM and 500 mM imidazole. Yield: 0.5 mg/L.

Subsequent LC MS analysis revealed successful cleavage of the N’ terminal methionine.However, a concomitant increase in mass of 12 26 Da (e.g. + 12) was observed. This massincrease is the result of thiazolidine ring formation (five or six membered) throughcondensation of the cysteine sulfhydryl group and the N’ terminal amino group withaldehydes (Figure 3 3A) such as formaldehyde (+ 12).37 This reaction of N’ Cys is wellcharacterized and has also found use in various applications, such as prodrugs,38 peptideligation,39 peptide purification,40 and N’ terminal protection of proteins.41 In the latter case,

CHAPTER THREE

56

the protective group is normally cleaved with methoxyamine, reverting the N’ Cys back intoits unprotected form as required for EPL (Figure 3 3A). This deprotection of the N’ terminusopening of thiazolidine ring was also successful in case of the recombinantly expressed

protein for NCL (Figure 3 3B).

NH2

HS

O

R

O

HHS

NH

O

RH2N

O

NH2

HS

O

R125 mM, pH 5.5

4 °C, 12h

Figure 3 3 | Protection N’ Cys as the thiazolidine ring and subsequent deprotection usingmethoxyamine(A) Reaction scheme for the protection and deprotection of N’ Cys; (B) liquid chromatography–mass spectrometry (LC MS) identifies the formation of the N’ terminal thiazolidine ring(A, + 12); (C) incubation with 125 mM methoxyamine hydrochloride in 500 mM acetate buffer at4 °C for 12 h (pH 5.5) reverts back to the free N’ terminus.

3.2.2.Generating chemically synthesized thioester peptide for EPL

The peptide for the EPL was synthesized by solid phase peptide synthesis (SPPS) usingFmoc protected amino acids on a Prelude peptide synthesizer. The coupling of the firstglycine to the linker on the resin with 2 (6 Chloro 1H benzotriazole 1 yl) 1,1,3,3tetramethylaminium hexafluorophos phate (HCTU) was very slow. To prevent incompletecoupling, the loading step of the first amino acid was repeated four times with the morereactive coupling reagent 2 (7 Aza 1H benzotriazole 1 yl) 1,1,3,3 tetramethyluroniumhexafluorophosphate (HATU). The subsequent residues, including N acetyl L lysine, weredoubly coupled with a subsequent N acetyl capping step. The last amino acid wasintroduced with an N’ terminal Boc protective group (Figure 3 4). Further details can befound in 3.4 Experimental: Solid Phase Peptide Synthesis & Purification.

B C

A


57

Figure 3 4 | Synthesis of the thioester peptide using the Dawson methodPeptide is elongated on a Dawson resin featuring a diaminobenzoyl (Dbz) linker usingstandard fluorenylmethyloxycarbonyl based synthesis (Fmoc). The last residue was coupled witha tert butyloxycarbonyl (Boc) protected N terminus. Activation of the Dbz group led to thecorresponding N acyl benzimidazolinone (Nbz) group, which was replaced by a thiol resulting ina C’ terminal thioester.35 R, amino acid side chains; R’SH, thiol additive. Further details can befound in 3.4 Experimental: Solid Phase Peptide Synthesis & Purification.

LC MS analysis of the cleaved crude peptides confirmed that the nonacetylated (KK Nbz),the lysine mono acetylated (K*K Nbz and KK* Nbz) and the lysine doubly acetylatedpeptide (K*K* Nbz) had been successfully formed (Figure 3 5A, crude). The peptides werethen purified by reverse phase HPLC. Due to poor separation of the highly polar peptidesvia standard C18 reversed phase based purification methods, the purification was insteadperformed using a hydrophilic interaction liquid chromatography (HILIC) column. Althoughthe reasoning behind peptide retention on the HILIC column is not yet fully understood, it isthought to be due to a combination of hydrophilic interactions and ion exchange. Theinteraction with hydrophilic compounds is facilitated by the aqueous layer which is formedon the silica surface covered with cross linked diol functional groups. In contrast to standardreverse phase chromatography, the strongest mobile phase has the highest waterconcentration, while the weakest has a higher concentration of the hydrophobic organicsolvent, acetonitrile (ACN).42 The purity of the peptides obtained after HILIC purificationwas significantly higher compared to standard RP chromatography. As an example, thepeptide KK* was isolated in sufficient purity for use in the optimization of the EPL reactionconditions (Figure 3 5).

CHAPTER THREE

58

Figure 3 5 | Purification of Dbz peptides using the HILIC columnHigh performance liquid chromatography (HPLC) monitoring purification of peptides. Total ionchromatogram (top, TIC), photodiode array (middle, PDA), and corresponding mass spectrum(MS, bottom) (A) after SPPS on standard RP octadecyl carbon (C18) chain bonded silica column(shallow gradient 1 10% ACN), and (B) after purification on analytical hydrophilic interactionchromatography (HILIC) column (inverse gradient, 70 10% ACN).

3.2.3.Test ligations using EPL with the model protein C YFP

In light of the low expression yield of C ER /S282C the ability of the chemically synthesizedpeptides to undergo protein ligation was first explored using a yellow fluorescent proteinwith an N’ Cys (C YFP) as a model protein. This protein showed improved stabilitycompared to the ER mutant (C ER /S282C) intended for NCL.43 A portion of the test ligationsfocused on the choice and concentration of the thiol additive, which was used to catalyze insitu transthioesterification during the ligation reaction. Furthermore, both the proteinpeptide ratio and the total concentration needed optimization, since they are known toinfluence ligation efficiency. As thiol catalyst, 4 mercaptophenylacetic acid (MPAA),thiophenol and 2 mercaptoethanesulfonate (MESNA) were used with protein:peptide ratiosranging from 1:10 to 1:33. The protein concentration varied between 5 116 M, while thepeptide varied from 0.1 1.7 mM (Table 3 1). Tris(2 carboxyethyl)phosphine) (TCEP) wasused as reducing agent (Table 3 1).

A BCrude C18 1-10% ACN HILIC purified 70-10% ACN

TIC

PDA

TIC

PDA

MS MS


59

Table 3 1 | Optimization of EPL of KK* with C YFP

reactionprotein[ M]

peptide[mM]

pro:pepratio

thiol catalyst TCEPvolume[ L]

MPPA 30 1.0 1:33I. 300mMMPAAII. 200 mMMPAA II. 20mM

150

MPPALOW

5 0.1 1:20I. 100mMMPAAII. 50 mMMPAA II. 20mM

150

MESNA 30 1.0 1:33I. 500mMMESNAII. 250 mMMESNA II. 20mM

150

MESNAHIGH

116 1.7 1:10II. 131 mMMESNA

I. 3.4 mMII. 1.7 mM

190

All reactions are performed in 0.2 M phoshate buffer, 10% (v/v) glycerol, pH 7.0 with C YFP and KK* thioester peptide.Ligation reaction: (I.) 1 h preincubation and (II.) main ligation reaction of 12 and 24 h, respectively.

The use of 2% (v/v) thiophenol resulted in a high degree of protein precipitation. However,in all other reactions protein instability was not an issue. A moderate level of ligationefficiency was observed, with a maximum conversion of 30% at high concentration of proteinand peptide (1:33 ratio) in combination with a high MPPA concentration (Table 3 1, reaction:MPPA) after 24 h reaction time (Figure 3 6A).

Figure 3 6 | Ligation efficiency of EPL of KK* and C YFP under different reaction conditionsMPPA and MESNA (protein:peptide: 1:33) both revealed moderate ligation, though marginallymore for MPAA. SDS PAGE analysis identified both unligated and ligated protein. Ligation:efficiency of ligation reaction is determined by integration of UV peak (photodiode array, PDA)of LC/MS spectrum.

Unfortunately, transfer of these ligation conditions to the recombinantly expressed CER /S282C construct resulted in protein precipitation and no evidence of successful ligation. Ingeneral, the EPL reaction has a number of limitations. Apart from the fact that thesite specific incorporation of chemically synthesized peptides can only be achieved close tothe termini of the protein, the ligation site must also be solvent accessible; thus denaturingconditions are in some cases required.44 However, in this study, the site of ligation site occursat the terminus of the 30 residue long HR. Thus accessibility of the cysteine residue shouldnot be an issue in this case. Most EPLs performed in separate literature studies wereperformed at the C’ terminus, meaning a synthetic peptide, possessing a N’ Cys, was ligatedto the C’ terminus of an expressed protein exhibiting a thioester.45 N’ terminal ligation,

CHAPTER THREE

60

however, requires a thioester peptide, which is sensitive to hydrolysis at pH > 7 (the optimalligation condition is pH 7). Thus, although the ligation rate of this bimolecular reaction isincreased by higher concentrations of the components, the ligation is still slow, andconsequently hydrolysis almost always accompanies the desired reaction.44 Analysis of KK*by means of LC MS analysis also identified hydrolyzed product over time. This confirms thatthe thioester required for NCL is not very stable.Additionally, the identity of the C’ terminal amino acid of the thioester peptide can

influence the rate of the EPL reaction. Although, the compatibility and efficiency of allproteinogenic amino acids could be shown, less hindered residues, including glycine andalanine resulted in a faster ligation.46 By contrast, valine, isoleucine, and proline werediscovered to react at a significantly slower rate, while aspartic acid and glutamic acid wereless effective due to side product formation.47 In our case, however, a glycine is present in theXaa Cys position which means that the reduced ligation efficiency cannot be explained byunfavored residues at the ligation site.The nature of the thiol leaving group is another factor that influences the reaction rate. In

the NCL the transthioesterification with the sulfhydryl group of the side chain of the N’ Cysis the rate limiting step. Therefore, exogenous thiols are added to NCL reactions to increaseligation kinetics by facilitating the in situ formation of more active thioesters.48 Common thioladditives are thiophenol or a benzyl mercaptan/thiophenol mix for peptide thioesters orMESNA for recombinant protein thioesters.49 A more recent, intense study investigating a setof 14 different sulfhydryl catalysts for NCL discovered aryl thiols, such as4 mercaptophenylacetic acid (MPAA), as highly effective catalysts.50 However, in this presentwork the use of MPAA instead of MESNA did not improve the ligation efficiencysignificantly, but instead resulted in a maximal ligation of 30% in case of the model proteinC YFP and no evidence of successful ligation in the case ofN’ ER /S282C.Another potential problem during the NCL reaction can be the formation of an unreactive

disulfide linked dimer via two N’ Cys. For this system, this oxidation reaction would lead toa covalent linkage of the N’ termini of two proteins, which would make the EPL reactionimpossible and could favor protein aggregation. During the EPL reaction, however, TCEP isadded as a reducing reagent preventing the thiol of the N’ Cys from oxidation.51 Disulfidebond formation might also occur after N’ terminal methionine cleavage in the cell. Whilepossible even under reducing conditions in the cytoplasm52 – by lowering the pKa of cysteinedue to charge interactions with adjacent residues53 – and close proximity of cysteines due toextension of the N’ terminus, it is less unlikely to happen in the cell. However, the fact thatN’ Cys is located at the terminus of a 25 residue long linker might be connected with the lowyield of the expression and high propensity of precipitation observed for this proteinconstruct.54 To study the impact of the length of the linker in more detail, constructs withstep wise elongation towards the HR were generated (see 3.2.4).


61

3.2.4.N’ terminal elongation of the ER LBD towards the ER HR

The influence of extending the N’ terminus of the ER LBD towards the HR was anotherpart of the investigation of the HR. This 60 residue long region connects the DNA bindingdomain (DBD) with the LBD and is considered to be flexible and unstructured.55,56 However,PTMs in this region of the receptor modulate transcriptional activity,57 receptordegradation,58 hormone sensitivity,59 functional synergy between AF 1 and AF 2,55 and DNAbinding.60 To gain more insight into the structure of the HR and its influence on the LBD, astep wise elongation of the LBD was performed. Thus, the LBD construct ER 309LBD553 FXaHis6 in pET44a was extended towards the HR by insertional mutagenesis. Five differentconstructs were thus obtained (Table 3 2):

Table 3 2 | Different elongated protein constructs

construct 1st position extension LBD (aa) domains total size (Da)LBD309 309 + 0 full length LBD 29,530INT294 294 +15 intermediate HR and LBD 31,330INT272 272 + 37 intermediate HR and LBD 33,572INT262 262 + 47 intermediate HR and LBD 34,863HR252 252 + 57 full length HR and LBD 35,988

According to the classification system used by the protein knowledgebase UniProtKB –which attributes 60 amino acids from 251 to 310 to the human ER HR (NR3A1)61 – LBD309(29,530 Da) is the full length LBD, HR252 (35,988 Da) is the full length LBD and HR andINT294, INT272, and INT262 are intermediate constructs (INT). Expression of theseconstructs in the presence of the natural ER full agonist estradiol (E2) was successful,resulting in good yields of protein (> 20 mg/mL) in the case of the LBD309 and INT294, andand moderate yields (around 5 mg/mL) in the case of INT272, INT262, and HR252,respectively (Figure 3 7).

Figure 3 7 | Protein expression and purification of HR252, INT262/272/294, and LBD309SDS PAGE gel with 100 mM (100), 250 mM (250), and 500 mM imidazole (500) elution fractions ofprotein expression and subsequent His6 tag purification with Ni NTA of ER LBD with differentHR extensions in the presence of the endogenous ligand, estradiol (E2).

309252 262 272 294

CHAPTER THREE

62

The protein expression without E2, however, resulted in low yields and a high degree ofimpurities, of which the shortest constructs – LBD309 and INT294 – yielded the purestproducts. Nevertheless, the yield was worse than had been initially expected compared toprotein expression in the presence of E2. (Figure 3 8).

Figure 3 8 | Ligand free protein expression and purification of HR252, INT’s, and LBD309SDS PAGE gel with 100 mM (100), 250 mM (250), and 500 mM imidazole (500) elution fractions ofprotein expression and subsequent His6 tag purification with Ni NTA of ER LBD with differentHR extensions in the absence of the endogenous ligand, estradiol (E2).

3.2.5. Influence of hinge elongation on protein stability and coactivator binding

The stability of elongated proteins LBD309, INT294, INT262, and HR252 was investigatedby circular dichroism (CD) spectroscopy. Only minor differences were observed for themelting temperatures of the four constructs (Tm: HR252 < LBD309 < INT294 < INT262,Figure 3 9A). In agreement with this finding, the spectrum of the mean residue ellipticity forthe four constructs afforded a similar pattern with similar 222 nm to 208 nm ratios (>1.0,Figure 3 9B). However, the results also clearly showed differences in mean residue ellipticity(MRE), despite all samples having been measured at 7.5 M concentrations. Interestingly, thelowest MRE were discovered for the longer protein constructs, while LBD309 had the highest(Figure 3 9B). The only explanation for this occurrence was due to precipitation during thepreparation procedure. Thus, it seems that large extensions towards the HR leads to a highertendency of the protein to precipitate. However, unfolding studies did not confirm this trendin the form of varying temperature dependent stabilities for the different protein constructs(Figure 3 9).

309252 262 272 294


63

Figure 3 9 | Unfolding curves and the mean residue ellipticity of the different HR constructsCircular dichroism based (CD) measurements: (A) the bottom asymptote of the mean residueellipticity at 222 nm at varying temperatures was set as folded protein, meaning 0% unfolding.The top asymptote at high temperatures was assumed as unfolded protein, thus 100% unfolding.Plotting temperature x against unfolding y led to the corresponding melting temperature Tm ofeach protein. (B) Full CD spectra with mean residue ellipticity at 20 °C using 7.5 M of theprotein in 10 mM phosphate buffer (pH 7.5). 222 nm/208 nm ratio as measure for the tertiarystructure.62

The ability of the elongated constructs to recruit coactivators (evaluated via interactionwith fluorescein labeled SRC1 Box2) was tested in a polarization assay. While the extensionfrom LBD309 via INT294 through to INT272 did not influence the binding properties(Kd = 0.36 0.41 M) the 10 residue step from INT272 to INT262 had a significant impact:namely, an increase in the dissociation constant Kd by one order of magnitude(Kd = 0.41 M 5.5 M). Although binding affinity improved for the HR252 construct(Kd = 1.2 M) it was still significantly lower than the three shorter constructs (Figure 3 10).

Figure 3 10 | Influence of HR elongation on coactivator recruitmentPolarization assay of different HR constructs. (A) Sigmoid curves from protein dilution series and(B) the corresponding Kd values.

B

A B

A

CHAPTER THREE

64

In summary, the extension of the LBD up to half way into the HR (272HR LBD553: INT272,INT294, LBD309) did not significantly influence receptor stability and its binding affinity toLXXLL motif. However, further extension into the HR (252HR LBD553) lowered the ability ofthe protein to recruit coactivator peptides. Although the CD spectrum (mean residueellipticity) of the long proteins HR252 and INT262 suggested lower receptor stability(through an increased propensity to precipitate) due to lowered concentration compared toINT294 and LBD309, temperature induced denaturation studies did not confirm this.Interestingly, studies on peptides based on the sequence of the HR revealed a higherpropensity to form an helix in the region close to the LBD (272HR310) compared to the regionclose to the DBD (252HR271), including Lys266 and Lys268 (chapter 2 and unpublished data inthe group). In both cases, however, the helical character could be improved upon theintroduction of PTMs. Thus, it seems to be that the ER HR can be divided into two regions,namely from residue 310 to 372 (272HR310) located close to the LBD, featuring moderatehelicity and a conserved RXKK motif, and from 271 to 252 (252HR271) close to the DBD,

being less helical and with possible PPII helix and two posttranslational modified lysines(chapter 2). Interestingly, in a reporter gene assay based study, acetylation of Lys266 andLys268 resulted in the stimulation of ligand dependent activity, whereas acetylation ofLys302 and Lys303 led to a diminishment of agonist response.63 This again supports thehypothesis of the presence of two major regions in the HR with different – or even opposing– outcomes upon the same PTM, such as lysine acetylation.

3.3. Conclusion

For EPL, both ligation partners were successfully synthesized. The recombinantlyexpressed protein construct C ER /S282C could be made. However, its synthesis required anadditional ring opening step to enable access to the protein with the N’ Cys. The peptides(KK Nbz, K*K Nbz, KK* Nbz, K*K* Nbz) possessing both site specific acetylations oflysines and a C’ terminal thioester were synthesized by solid phase peptide synthesis. Whiletest ligations of the peptides to a model protein could be optimized as far as 30% ligationefficiency, transfer of the same ligation conditions to ER was not successful. Possiblepeptide hydrolysis and mainly protein instability (manifest as an increased tendency for theprotein to precipitate) were the major issues in this case. The studies concerning theenlargement of the ER LBD towards the HR revealed that expression as far as position 272in the presence of E2 gave good expression yields (INT272, INT294, LBD309). The meltingpoint, reflecting the thermal stability of the protein, was not influenced upon furtherextension. However, the expression yields diminished and binding studies with coactivatorpeptides identified a significant decrease in affinity (INT252, INT262). These results, togetherwith studies on peptides based on the HR sequence, suggested the presence of two distinctregions of the HR with different structure and function upon posttranslational modification.


65

3.4. Experimental

GeneralIf not stated otherwise the following contents, descriptions, protocols, and distributer are valid:chemicals were ordered from SIGMA ALDRICH. Proteins were handled at 4 °C. Standard culturingmedium is lysogeny broth (LB) medium (10 g peptone, 10 g NaCl, 5 g yeast, 1 l water, autoclaved).SRC1 Box2 or short S1B2 is the 20 amino acid long sequence of the nuclear receptor coactivator 1(Ncoa1692 711) with the sequence SLTERHKILHRLLQEGSPSD, including a LxxLL motif.

Protein Expression & Purification

Table 3 3 | Used constructs for protein expression

construct vector tag protein originER full length pCDNA3 hER 1 595 GIFT FROM ROBMICHALIDES

ER LBD309 pET44a FXa His6 (C ter.) hER LBD309 553 CLONING FROM ER full length

ER INT294 pET44a FXa His6 (C ter.) hER HR LBD294 553 CLONING FROM ER full length



ER HR252 pET44a FXa His6 (C ter.) hER HR LBD252 553 CLONING FROM ER full length

Plasmids with the desired protein construct (Table 3 3) were transformed into E. coli BL21 cells.Emerging colonies were cultured over night in 25 ml LB medium with 100 g/ml ampicillin at 37 °C.This preculture was added to 2 l phosphate buffered, content rich terrific broth (TB) medium (12 gpeptone, 24 g yeast, 4 ml glycerol, 0.17 M KH2PO4, 0.72 M K2HPO4, 1 l water, autoclaved) withampicillin and incubated at 37 °C until an optical density at 600 nm (OD600) of 0.6 1.0. Addition of100 M Isopropyl D 1 thiogalactopyranoside (IPTG) induced expression of the protein of interest.For stabilization 10 M of ligand (ER, estradiol) was added. After incubation for 20 h at 15 °C the cellswere harvested by centrifugation (15 min at 7,000 rpm) and the resulting pellet was stored at 80 °Cuntil further usage.If not stated otherwise, the protein purification was performed at 4 °C. The pellet was resuspended inlysis buffer (LyB 1: 1x PBS tablet, CALBIOCHEM, 370 mM NaCl, 10% (v/v) glycerol, 0.1 mM TCEP,1 mM PMSF, 1 g/ml DNAseI, 10 M ligand, pH 8.0, His6 tag: additional 40 mM imidazole) and lyzedwith the Emulsi Flex C3 homogenizer (2 passes of 150,000 kPa, AVESTIN INC.). After anothercentrifugation step (40 min, 20,000 rpm) the supernatant was provided on the respective affinitychromatography column (His6 tag: His Bind Resin, 3 ml, NOVAGEN), washed with wash buffer (1x PBStablet, CALBIOCHEM, 370 mM NaCl, 40 mM imidazole, 10% (v/v) glycerol, 0.1 mM TCEP, 10 M ligand,pH 8.0) and eluted with competing molecules (WaB 1 plus 100 mM / 250 mM / 500 mM imidazole).Fractions were analyzed with SDS PAGE and optional with LC MS.The N’ terminal thiazolidine ring that was formed between the cysteine side chain and the freeN’ terminus of the protein for EPL (+ 11 21) could be opened with treatment of 125 mMmethoxyamine in 500 mM acetate buffer at 4 °C for 24 h (pH 5.5). LC MS confirmed the successfulreaction.

CHAPTER THREE

66

Solid Phase Peptide Synthesis & Purification35

The synthesis of all peptides was performed on solid support using the Fmoc strategy on a Preludepeptide synthesizer (PROTEIN TECHNOLOGIES INC.). If not stated otherwise, 200 mol (1 eq.) of DawsonDbz AM resin (0.43 mmol/g, 100 200 mesh, NOVABIOCHEM) was swollen for 30 min in 5 mlN methylpyrrolidone (NMP). Washing was performed with NMP or dichloromethane (DCM),deprotection of the amino acids (AA, 0.2 M solution in NMP) was performed with piperidine (Dep.,20% (v/v) solution in NMP), activated with 2 (7 Aza 1H benzotriazole 1 yl) 1,1,3,3tetramethyluronium hexafluorophosphate (Act1stAA, HATU, 0.4 M solution in NMP) or 2 (6 Chloro 1Hbenzotriazole 1 yl) 1,1,3,3 tetramethylaminium hexafluorophosphate (Act., HCTU, 0.4 M solution inNMP) and alkalized with N,N diisopropylethylamine (Base, DIPEA, 1.6 M in NMP). The loading of thefirst amino acid was performed in a four times coupling: 1 ml Dep. (5 min), 3x[2 ml NMP, 2 ml AA,1 ml Act1stAA, 1 ml Base, 20 min], 1 ml NMP, 2 ml DCM. The following residues are double coupledwith a subsequent capping step (Cap., 20% (v/v) pyridine, 20% (v/v) acetic anhydride in NMP): 6 mlNMP, 4 ml Dep. (5 min), 2x[6 ml NMP, 4 ml AA, 2 ml Act., 2 ml Base, 20 min], 6 ml NMP, 3 ml Cap.(5 min), 3 ml NMP. The last amino acid was coupled BOC protected. After finishing the desiredpeptide sequence, the resin was washed 3 times alternately with dimethylformamid (DMF) anddichloromethane (DCM) and the diaminobenzoyl (Dbz) group was activated 1 h with 0.5 mmolp nitrophenyl chloroformate (p NPCF) resulting in the corresponding N acyl benzimidazolinone(Nbz). After washing with DCM, 0.5 M DIPEA in DMF was added for 30 min. After another washingstep with DMF and DCM, the side chain protection group and the cleavage of the peptide sequencefrom the resin were simultaneously achieved by the usage of mixture of trifluoroacetic acid,triisopropylsilane and water (TFA/TIS/H2O, 95/2.5/2.5) for 3 h. The free peptide was precipitated byadding it drop wise to ice cold DEE. A final centrifugation step (2.000 rpm, 10 min) with subsequentwashing (ice cold DEE), uptake in water and lyophilization results in the crude peptide. Purificationwas accomplished by HPLC on a Luna® 5 m HILIC 200 Å, LC Column 150 x 21.2 mm, AXIA™Packed (PHENOMENEX). Acetonitrile (ACN) was used as a apolar phase, adding different amounts ofan polar ammonium formate (10 mM, pH3.5) phase. A linear gradient (20 ml/min) was optimized foreach peptide to a 10% range, for instance 60 50% ACN. The purity of the peptide was determinate atan analytical LC MS using Luna® 5 m HILIC 200 Å, LC Column 100 x 2 mm (PHENOMENEX). Finally,the peptides were lyophilized and stored at 80 °C.


67

Fluorescence Polarization Assay64,65

Fluorescence polarization (FP) experiments were measured in black 384 well plates (PERKIN ELMER 384F) on a Safire2 (TECAN) plate reader. The final volume of one well was 30 l containing constantconcentrations of ligand (5 M) and Fluorescein labeled peptide (0.1 M, ER: NCoA 1/SRC1683 701, FLLTERHKILHRLLQEGSPSD). The protein was sequentially diluted (24 steps) in TR FRET coregulatorbuffer E (INVITROGEN). Briefly, in 96 well plates dilution series were prepared in 55 l and filled upwith to 110 l with a 2X master mix solution containing all non varying components. Finally threetimes 30 l was transferred to 384 well plates, centrifuged (1,000 rpm, 2 min) and incubated for 1 h at4 °C. Plates were measured 50 times at 30 °C (excitation 470 nm, emission 519 nm). Polarization (in[mP]) was plotted against the concentration of either the protein (FP), whereby each data pointrepresents an average of 3 experiments.The dissociation constant Kd of the protein peptide complex (FP) was calculated using equation 1:

where is the bottom asymptote, is the top asymptote, and the hill slope (steepness of thecurve).

Circular Dichroism spectroscopy66,67

Far UV Circular Dichroism (CD) spectroscopy measurements were performed under a constantnitrogen flow at 20 °C using 7.5 M of the protein in 10 mM phosphate buffer (pH 7.5), if not statedotherwise. As a baseline a protein free buffer was measured under identical conditions. The spectrumwas recorded from 250 nm to 185 nm using a JASCO 815 spectrometer. Quartz cuvettes with pathlengths of 1 mm (HELLMA ANALYTICS) were employed with a data pitch of 0.5 in a continuous mode, ascan speed of 20 nm/min with a response time of 2 s and a bandwidth of 0.5 nm. The graphs arerepresenting an average of five scans. The observed ellipticity (degrees, in [mdeg]) was converted tomean residue ellipticity (in [deg cm2/dmol]) using DICHROWEB.66 Information about helicalpropensity in semi quantitative manner could be shown by the ratio of the characteristic wavelengthsat 208 nm and 222 nm: .For the unfolding studies the mean residue ellipticity at 222 nm was used to detect unfolding of theprotein. The bottom asymptote at low temperature was set as folded protein, meaning 0% unfolding.By contrast, the top asymptote at high temperatures – no further change of mean residue ellipticity at222 nm upon increase of temperature – was assumed as unfolded protein, thus 100% unfolding.Plotting temperature x against unfolding y led to corresponding melting temperature Tm of eachprotein.

Expressed Protein LigationAll optimization were performed with recombinantly expressed CYFP (provided by Ralph Bosmann)and KK* Nbz peptide in 0.2 M phoshate buffer, including 10% (v/v) glycerol at pH 7.0 at roomtemperature for 12h and 24h, respectively. Protein , peptide , thiol additive , and TCEP concentrationcan be seen in Table 3 1. The ER ligation trials were performed with MPPA and MESNA conditions(Table 3 1,MPPA/MESNA).

CHAPTER THREE

68

3.5. References

1. J. A. Brannigan and A. J. Wilkinson, Nat Rev Mol Cell Biol, 2002, 3, 964–970.2. P. E. Dawson, T. W. Muir, I. Clark Lewis, and S. B. Kent, Science, 1994, 266, 776–779.3. T. S. Young and P. G. Schultz, J Biol Chem, 2010, 285, 11039–11044.4. C. P. R. Hackenberger and D. Schwarzer, Angew Chem Int Ed, 2008, 47, 10030–10074.5. T. Wieland, E. Bokelmann, L. Bauer, H. U. Lang, and H. Lau, Liebigs Ann Chem, 1953, 583, 129–149.6. S. B. H. Kent, Chem Soc Rev, 2009, 38, 338.7. J. A. Camarero, G. J. Cotton, A. Adeva, and T. W. Muir, J Pept Res, 1998, 51, 303–316.8. V. Muralidharan and T. W. Muir, Nat Methods, 2006, 3, 429–438.9. R. R. Flavell and T. W. Muir, Acc Chem Res, 2009, 42, 107–116.10. J. A. Camarero, Y. Kwon, and M. A. Coleman, J Am Chem Soc, 2004, 126, 14730–14731.11. J. A. Camarero and T. W. Muir, J Am Chem Soc, 1999, 121, 5597–5598.12. A. Romanelli, A. Shekhtman, D. Cowburn, and T. W. Muir, Proc Natl Acad Sci USA, 2004, 101,

6397–6402.13. F. I. Valiyaveetil, R. MacKinnon, and T. W. Muir, J Am Chem Soc, 2002, 124, 9113–9120.14. L. M. Szewczuk, M. K. Tarrant, V. Sample, W. J. Drury 3rd, J. Zhang, and P. A. Cole, Biochemistry,

2008, 47, 10407–10419.15. L. Brunsveld, J. Kuhlmann, K. Alexandrov, A. Wittinghofer, R. S. Goody, and H. Waldmann,

Angew Chem Int Ed, 2006, 45, 6622–6646.16. M. Shogren Knaak, H. Ishii, J. M. Sun, M. J. Pazin, J. R. Davie, and C. L. Peterson, Science, 2006,

311, 844–847.17. S. Möcklinghoff, R. Rose, M. Carraz, A. Visser, C. Ottmann, and L. Brunsveld, ChemBioChem, 2010,

11, 2251–2254.18. D. A. Erlanson, M. Chytil, and G. L. Verdine, Chem Biol, 1996, 3, 981–991.19. T. J. Tolbert and C. H. Wong, Angew Chem Int Ed, 2002, 41, 2171–2174.20. D. Liu, R. Xu, K. Dutta, and D. Cowburn, FEBS Lett, 2008, 582, 1163–1167.21. R. E. Dalbey, M. O. Lively, S. Bron, and J. M. van Dijl, Protein Sci, 1997, 6, 1129–1138.22. M. Paetzel, R. E. Dalbey, and N. C. J. Strynadka, J Biol Chem, 2002, 277, 9512–9519.23. P. S. Hauser and R. O. Ryan, Protein Expr Purif, 2007, 54, 227–233.24. P. H. Hirel, M. J. Schmitter, P. Dessen, G. Fayat, and S. Blanquet, Proc Natl Acad Sci USA, 1989, 86,

8247–8251.25. J. A. Camarero, D. Fushman, D. Cowburn, and T. W. Muir, Bioorg Med Chem, 2001, 9, 2479–2484.26. T. C. Evans Jr, J. Benner, and M. Q. Xu, J Biol Chem, 1999, 274, 3923–3926.27. S. Mathys, T. C. Evans, I. C. Chute, H. Wu, S. Chong, J. Benner, X. Q. Liu, and M. Q. Xu, Gene,

1999, 231, 1–13.28. M. W. Southworth, K. Amaya, T. C. Evans, M. Q. Xu, and F. B. Perler, BioTechniques, 1999, 27, 110–

114, 116, 118–120.29. J. A. Camarero, B. J. Hackel, J. J. de Yoreo, and A. R. Mitchell, J Org Chem, 2004, 69, 4145–4151.30. J. Alsina, T. S. Yokum, F. Albericio, and G. Barany, J Org Chem, 1999, 64, 8761–8769.31. Y. Shin, K. A. Winans, B. J. Backes, S. B. H. Kent, J. A. Ellman, and C. R. Bertozzi, J Am Chem Soc,

1999, 121, 11684–11689.32. P. Botti, M. Villain, S. Manganiello, and H. Gaertner, Org Lett, 2004, 6, 4861–4864.33. Y. Ohta, S. Itoh, A. Shigenaga, S. Shintaku, N. Fujii, and A. Otaka, Org Lett, 2006, 8, 467–470.34. D. Swinnen and D. Hilvert, Org Lett, 2000, 2, 2439–2442.35. J. B. Blanco Canosa and P. E. Dawson, Angew Chem Int Ed, 2008, 47, 6851–6855.36. J. Tsuji, R. Nydza, E. Wolcott, E. Mannor, B. Moran, G. Hesson, T. Arvidson, K. Howe, R. Hayes,

M. Ramirez, and M. Way, BIOS, 2010, 81, 22–31.37. M. P. Schubert, J Biol Chem, 1936, 114, 341–350.38. B. H. Wilmore, P. B. Cassidy, R. L. Warters, and J. C. Roberts, J Med Chem, 2001, 44, 2661–2666.


69

39. C. F. Liu and J. P. Tam, Proc Natl Acad Sci U S A, 1994, 91, 6584–6588.40. M. Villain, J. Vizzavona, and K. Rose, Chemistry & Biology, 2001, 8, 673–679.41. D. Bang and S. B. H. Kent, Angew Chem Int Ed, 2004, 43, 2534–2538.42. B. Buszewski and S. Noga, Anal Bioanal Chem, 2012, 402, 231–247.43. O. V. Stepanenko, V. V. Verkhusha, I. M. Kuznetsova, V. N. Uversky, and K. K. Turoverov, Curr

Protein Pept Sci, 2008, 9, 338–369.44. M. R. Seyedsayamdost, C. S. Yee, and J. Stubbe, Nat Protoc, 2007, 2, 1225–1235.45. L. Berrade and J. A. Camarero, Cell Mol Life Sci, 2009, 66, 3909–3922.46. T. M. Hackeng, J. H. Griffin, and P. E. Dawson, PNAS, 1999, 96, 10068–10073.47. M. Villain, H. Gaertner, and P. Botti, Eur J Org Chem, 2003, 2003, 3267–3272.48. P. E. Dawson, M. J. Churchill, M. R. Ghadiri, and S. B. H. Kent, J Am Chem Soc, 1997, 119, 4325–

4329.49. T. W. Muir, Annu Rev Biochem, 2003, 72, 249–289.50. E. C. B. Johnson and S. B. H. Kent, J Am Chem Soc, 2006, 128, 6640–6646.51. J. A. Burns, J. C. Butler, J. Moran, and G. M. Whitesides, J Org Chem, 1991, 56, 2648–2650.52. C. S. Sevier and C. A. Kaiser, Nat Rev Mol Cell Biol, 2002, 3, 836–847.53. A. Rietsch and J. Beckwith, Annu Rev Genet, 1998, 32, 163–184.54. G. Bulaj, Biotechnol Adv, 2005, 23, 87–92.55. W. Zwart, R. de Leeuw, M. Rondaij, J. Neefjes, M. A. Mancini, and R. Michalides, J Cell Sci, 2010,

123, 1253–1261.56. A. Haelens, T. Tanner, S. Denayer, L. Callewaert, and F. Claessens, Cancer Res, 2007, 67, 4514–4523.57. S. Sentis, M. L. Romancer, C. Bianchin, M. C. Rostan, and L. Corbo,Mol Endocrinol, 2005, 19, 2671–

2684.58. N. B. Berry, M. Fan, and K. P. Nephew,Mol Endocrinol, 2008, 22, 1535–1551.59. C. Wang, M. Fu, R. H. Angeletti, L. Siconolfi Baez, A. T. Reutens, C. Albanese, M. P. Lisanti, B. S.

Katzenellenbogen, S. Kato, T. Hopp, S. A. W. Fuqua, G. N. Lopez, P. J. Kushner, and R. G. Pestell,J Biol Chem, 2001, 276, 18375–18383.

60. M. Y. Kim, E. M. Woo, Y. T. E. Chong, D. R. Homenko, and W. L. Kraus, Mol Endocrinol, 2006, 20,1479–1493.

61. The UniProt Consortium, Nucleic Acids Res, 2011, 40, D71–D75.62. N. J. Greenfield, Nat Protoc, 2006, 1, 2527–2535.63. R. A. Wang, A. Mazumdar, R. K. Vadlamudi, and R. Kumar, EMBO J, 2002, 21, 5437–5447.64. P. Wu, M. Brasseur, and U. Schindler, Anal Biochem, 1997, 249, 29–36.65. F. Perrin, J Phys Radium, 1926, 7, 390–401.66. L. Whitmore and B. A. Wallace, Nucleic Acids Res, 2004, 32, W668–W673.67. L. Whitmore and B. A. Wallace, Biopolymers, 2008, 89, 392–400.

CHAPTER FOUR 4 Screening the Estrogen Receptor Coactivator

Interaction

Abstract. Ribosome display was used to effectively screen the surface of the estrogenreceptor (ER) ligand binding domain (LBD) for novel natural peptide binders. Whereasearlier rounds of ribosomal enrichment witnessed the expected emergence of the hallmarkLXXLL motif subsequent rounds led to the identification of a more mature PXLXXLLXXPconsensus, which could be validated by both biochemical and cellular methods. Molecularmodeling (MD) and X ray crystallography studies clearly defined a specific role for theflanking prolines as helix breakers, which prime the helix length for optimal interaction withthe surface charge clamp. Furthermore, the conformational constraints imposed by theprolines on adjacent immediate flanking amino acid residues are believed to determinereceptor subtype selectivity and binding affinity through the precise orientation of side chainfunctionality at the ER surface. These findings resulted in potent peptide inhibitors of theER coactivator interaction based alone on natural sequences and provide a structuralrationale for the function of prolines in natural coactivator proteins. Finally, this workrepresents a fundamental re evaluation of the NR coactivator interaction and thus sets new,minimal, structural parameters based on linear sequences of proteinogenic amino acids.Moreover it also provides insights for peptide derived tools and more druggable peptideagents.

Sascha Fuchs, Hoang D. Nguyen, Trang T.P. Phan, Matthew F. Burton, Lidia Nieto,Ingrid J. de Vries van Leeuwen, Andrea Schmidt, Monireh Goodarzifard, Stijn M. Agten,

Rolf Rose, Christian Ottmann, Lech Gustav Milroy, and Luc Brunsveld, submitted

Chapter Four

72

4.1. Introduction

The standard approach to treat diseases caused by the misregulation of nuclear receptors(NR) is the use of small lipophilic drug molecules which modulate NR mediated genetranscription at the ligand binding pocket (LBP) of the ligand binding domain (LBD).Recently, in the case of the estrogen receptor (ER; , NR3A1; , NR3A2), compounds withreduced toxicity have been developed, namely selective estrogen receptor modulators(SERMs).1 One example is tamoxifen which achieves improved cell and tissue selectivitythrough a more distinct recruitment of coactivator proteins.2–4 While clearly of significantbenefit, classical approaches targeting the LBP suffer from a number of drawbacks, includingendocrine resistance,5 and a general lack of selectivity at the level of coactivator recruitment,resulting in an urgent need for alternative approaches to modulate ER mediated genetranscription.6

A promising alternative strategy is direct inhibition of the protein protein interactionbetween ER and its coactivators.7 In general, agonist binding such as the endogenousligand estradiol (E2) in case of ER at the LBP of the LBD induces a conformational change,most prominent in the helix 12 (H12) region, which favors binding of an helical LXXLLconsensus motif present in various NR coactivator.8,9 A defining feature of this interaction isthe charge clamp (with a positive charged lysine and negative charged glutamic acid), whichis formed at the interface between H12 and other parts of the activation function 2 (AF 2).This clamp defines the maximal length of an helical peptide for optimal coactivatorbinding and offers therefore a unique opportunity for the development of selectiveinhibitors.10 Seminal work involving the screening of large LXXLL derived peptide librariesby phage display led to the identification of potent and selective ER coactivator bindinginhibitors (CBI).11 However, issues of poor cell permeability and low metabolic stabilitytypically complicate the translation of peptide hits into useful drug therapies.12 This hasmotivated to recent efforts to develop small lipophilic drug like molecules which target thecharge clamp or mimic the leucine side chains of the coactivator helix, via computationaldesign or using high throughput screening (HTS).13–18 Modified peptide inhibitors have alsobeen investigated with improved binding affinity, ER subtype selectivity and proteolyticstability.19–21

In recent years peptide drug discovery has matured and a number of different strategies tostabilize peptides have emerged,22 including macrocylization via disulfide bond formation,19

macrolactamization,20 hydrocarbon stapling,23 or the incorporation of hydrogen bondingsurrogates,24 which preorganize the helix through stabilization of its structure in thesurface bound state. Nonetheless, the drawbacks of such approaches are the necessity forsynthetic operations to upgrade amino acid building blocks to facilitate macrocyclization andthe unpredictable effects that such modifications might have on binding. Furthermore,advances in this field have built only on prior knowledge of natural coactivators or peptide

Screening the Estrogen Receptor Coactivator Interaction

73

sequences derived from phage display screening, and are therefore restricted to the inherentdiversity of the systems at hand. In light of this, we wondered whether a consensus had yetto be reached on the minimal structural requirements for potent inhibition of theER coactivator interaction using only natural amino acids. The acquisition of suchknowledge would greatly depend on the evaluation of vast, more structurally diversepeptide libraries than had previously been tested. Indeed the NR coactivator interaction wasfirst evaluated using phage display,25 and it was thanks to the efficiency of this method thatthe LXXLL motif was discovered as a preferential binder to ER and other NR surfaces.26,27

Nonetheless, the usage of an alternative screening approach, which samples from a morediverse peptide library, might reveal optimized extended LXXLL consensus motifs or evenalternative sequences with improved binding profiles. Next to the well established phagedisplay,25 also the yeast two hybrid system,30 plasmid ,31 and cell surface display areavailable.32,33 But since they all are limited by their intrinsic transformation efficiency, thecomplexity of these libraries are at most 109 unique peptide entities. Cell free in vitromethodologies, including ribosome display (Figure 4 1),34–36 however, are resolving thislimitation resulting in a random peptide library with high complexity and up to 1014

independent members.37 This advantage of that technique, first described by Mattheakis et al.and further developed by Plückthun et al.,28,29 could enable a closer examination of theflanking regions of the LXXLL region than had previously been possible.

Chapter Four

74

Figure 4 1 | Ribosome Display against immobilized ER LBDThe DNA library with the randomized region, resulting in SLTARXXXXXRXXXXXSPSD peptideswas expressed in vitro. The library of ribosomal complexes, consisting of the expressed peptidetethered to the ribosome with unreleased mRNA, was screened against immobilized ER LBDand ER LBD, respectively. The geno and phenotype including library was used as a basis forreversed transcription resulting in an optimized DNA library with enhanced ER bindingproperties. Several rounds of selection resulted in an optimized DNA library that was analyzedby colony PCR, sequencing, and final conversion to analogue amino acids sequence.

4.2. Ribosome display screening and preliminary validation

Initial rounds of ribosomal enrichment led to the emergence of the LXXLL consensus motiffor both ER and ER subtypes. Further rounds of enrichment, however, resulted in a morerefined PXLXXLLXXP motif, which was particularly prominent in the case of ER (Figure4 2).


75

Figure 4 2 | Cluster analysis of different rounds of ribosome display screeningHeight of letters is representative of the abundance of an amino acid at certain positions of theamino acid sequence after ribosome display screening against the LBD of ER (left) and ER(right), respectively. Alignment and position labeling was made relative to the emerging LXXLLmotifs.38

Hit sequences were selected for a preliminary validation with SNAP tag based cellularcompetitive inhibition studies. These binding tests performed in mammalian U2OS cell linesidentified proline based peptide sequences 4 (PXLXXLLXXP), 5 (PXLXXHL) and 10(PXLXXLLXXP) as the most efficacious binders compared with the natural SRC1 Box2sequence (1).

Chapter Four

76

Figure 4 3 | Validation of peptide sequences identified by ribosome display screening(A) LXXLL aligned sequences of naturally derived NCoA1 Box 2 sequence, 1, and peptidesequences 2 10 identified after round eight of ribosome display. Comparison of bindingproperties of peptides 1 10 by Kd values (mean ± s.e.m.) obtained as mEGFP peptide fusions by abinding fluorescence polarization assay (B) or testing their luciferase activity (mean ± s.e.m.) in acellular competitive inhibition assay against ER (C) or ER (D), respectively.

Based on both the cluster analysis data (Figure 4 2) and results from the preliminarycellular screening (Figure 4 3), it was hypothesized that the presence of the prolines might beplaying a significant role in determining the improved binding efficacy of sequences 4, 5, and10 compared with the natural sequence (1). Therefore, 4, 5, and 10 were synthesized by solidphase peptide synthesis (SPPS) together with a series of rationally designed truncatedanalogues (Table 4 1) in order to probe the importance of the prolines in subsequentbiochemical and structural studies.

C D

BA HKILHRLLQEG (1)SSLLIRLLQLP (2)MPQLTRLLLSP (3)SPLLTRLLLQP (4)HPLLLRHLLQN (5)LLMSSRLITLL (6)LLQPNRLLSLL (7)LLLSSRLLSLL (8)LLSSPRQQTYH (9)HPLLMRLLHHP (10)


77

4.3. Exploration of the PXLXXLLXXP motif

All peptides were first measured in a competition fluorescence polarization assay usingfluorescein labeled 1 as the competing peptide (Figure 4 4A). The main focus during furtherinvestigations was on the subtype ER , where the emergence of the prolines was discoveredto be most prominent (Figure 4 2B). Overall, the peptides produced classical sigmoidalcurves and behaved as competitive inhibitors with Ki values in the range 25 nM to 75 M(Figure 4 4B D, Table 4 1).

Figure 4 4 | Fluorescence Depolarization Assay(A) General principle of the fluorescence depolarization assay based on the ER SRC1 Box2interaction. (B) Normalized sigmoid curves from the evaluation of 22 against other NRs.Fluorescein labeled competing peptides: ER , SRC1 Box2 (1); AR, FLETT 1; RXR , D22.Reference peptides: ER , 1; AR, AR NTD (26); RXR , phage display hit against ER and goodbinding properties for RXR (27).26 (C) Selected (1, 4, 5, 13, 22) sigmoid curves resulting from ERbinding studies. (D) Selected (1, 4, 5, 13, 22) sigmoid curves resulting from ER binding studies.IC50 and Ki values for both receptor subtypes in Table 4 1.

The ribosome sequences 4, 5, and 10, were found to bind more potently to ER (Ki < 50 nM)than the natural 1 sequence (Ki = 122 nM). Truncation of peptide sequences 4 and 10 resultedin a decrease in binding affinity (Table 4 1). N’ terminal truncation of 19 mer 4 yielded 15mer 11, which was 2.5 times less active than the parental peptide. By comparison,

A B

C D

Chapter Four

78

N’ terminal truncation of 19 mer 10 gave 15 mer 12, which was significantly less potent than10. Further truncation of 12 at the C’ terminus, yielded 12 mer 13, which was similarly(in)active as 12 (Ki = 3.6 M). This data would suggest that, at least for 4 and 10, thefar flanking regions of the LXXLL motif make an important contribution to binding, and thatthe presence of the proline residues alone is not sufficient for high potency binding.

Table 4 1 | Binding affinities of synthesized peptides

peptide sequence nameestrogen receptor estrogen receptor

IC50 ± s.e.m.b[ M]

Ki (±)b[ M]

IC50 (±)b[ M]

Ki (±)b[ M]

LTERHKILHRLLQEGSPSD 1 0.470 ± 0.059 0.122 ± 0.039 0.233 ± 0.045 < 0.040aLTARSPLLTRLLLQPSPSD 4 0.171 ± 0.009 < 0.050a 0.186 ± 0.007 < 0.040aLTARHPLLLRHLLQNSPSD 5 0.184 ± 0.044 < 0.050a 0.251 ± 0.009 < 0.040aLTARHPLLMRLLHHPSPSD 10 0.279 ± 0.028 < 0.050a 0.756 ± 0.022 0.185 ± 0.012

SPLLTRLLLQPSPSD 12 0.466 ± 0.040 0.119 ± 0.026 n.d.e n.d.fHPLLMRLLHHPSPSD 11 7.150 ± 0.850 4.622 ± 0.245 n.d.e n.d.fHPLLMRLLHHPS 13 5.381 ± 0.611 3.640 ± 0.210 3.839 ± 0.254 1.541 ± 0.111

AcEPILHRLLQKP 14 38.35 ± 17.92 25.66 ± 12.09 n.d.e n.d.eAcSPILHRLLQEP 15 n.b.d ( ) n.b.d ( ) n.d.e n.d.e

AcRHKPLHRLLQEPS 16 98.24 ± 12.52 66.04 ± 8.446 n.d.e n.d.eAcRHPILHRLLQPEG 17 2.832 ± 0.118 1.711 ± 0.080 n.d.e n.d.eAcRHPLHRLLQEAPS 18 112.0 ± 17.64 75.35 ± 11.89 n.d.e n.d.eRHPKILHRLLQPE 19 1.460 ± 0.098 0.787 ± 0.066 n.d.e n.d.eAcPILHRLLQEP 20 0.680 ± 0.102 0.262 ± 0.068 n.d.e n.d.eAcHPILHRLLQEP 21 0.395 ± 0.020 0.072 ± 0.013 0.964 ± 0.002 0.286 ± 0.001AcHPLLMRLLLSP 22 0.132 ± 0.021c 0.025 ± 0.015c 0.732 ± 0.029 0.185 ± 0.012AcLLMRLLLS 28 0.301 ± 0.050 < 0.050a 0.859 ± 0.034 0.240 ± 0.014HPLLMRLLLSP 23 0.289 ± 0.017 < 0.050a n.d.e n.d.e

AcHKILHRLLQEG 24 0.591 ± 0.120 0.203 ± 0.080 0.602 ± 0.029 0.128 ± 0.013AcKILHRLLQEG 25 0.695 ± 0.144 0.272 ± 0.096 n.d.e n.d.eAcHPLLMRLLLSP 22 against AR LBD 16.61 ± 7.489AcHPLLMRLLLSP 22 against RXR LBD 0.722 ± 0.196LXXLL motif highlighted in grey, flanking prolines underlined and bold; a, no exact value determinable due to assay

limitation, b, standard error of the mean; c, non standard conditions: 100 nM protein, d, no binding, e not determined; lasttwo rows against AR/RXR LBD, 22measuring in presence of AR and RXR LBD, respectively, instead of ER or ER LBD.

Interestingly though, insertion of the two proline residues into the 11 mer natural 1sequence at the same 2 and +3 positions relative to the LXXLL motif as observed in 4 and10 (Table 4 1, 21) resulted in a more potent binder (Ki = 72 nM) than the natural 11 mersequence, 24 (Ki = 203 nM). Furthermore, this short proline based peptide was more potentthan the natural 19 mer sequence 1 (Ki = 122 nM). Interestingly, an 11 mer peptide consistingof the most frequent amino acids derived from the cluster analysis for ER (Figure 4 2 &Table 4 1, 22) resulted in high affinity binder (Ki = 25 nM). Modifications to negativelycharged or polar residues adjacent to the proline residues (Table 1, 14 and 15) caused thepeptide binding affinity to diminish. Subsequent investigations into the importance of thepositioning of prolines relative to the LXXLL motif suggested that the proline at 2 was


79

optimally placed (Table 4 1, 16 and 18), while hinting a greater flexibility for the proline at +3in the C’ terminal region (Table 4 1, 17 and 19). Taken together, these data suggest that theprolines make a significant and important contribution to peptide binding at the ER surface.

4.4. Structural analysis of potent binders

Circular dichroism (CD) measurements were performed to determine the helicity of thepeptides (Table 4 2). Whereas in pure phosphate buffer all peptides measured were randomcoil, the addition of 30% (v/v) trifluoroethanol (TFE) – reported to stabilize intra molecularhydrogen bonds and therefore mimic the environment at the protein surface – resulted in anincrease in the percentage helicity for most of the peptides measured.40 This stronglysuggests that helix nucleation and stabilization depend on binding at the NR surface.

Table 4 2 | Circular dichroism measurements of selected peptides

peptide sequence name 222/ 208c CD spectraLTERHKILHRLLQEGSPSD 1a 0.85LTARSPLLTRLLLQPSPSD 4a 0.56LTARSPLLTRLLLQPSPSD 4b 0.25bLTARHPLLLRHLLQNSPSD 5a 0.79LTARHPLLMRLLHHPSPSD 10a 0.75

SPLLTRLLLQPSPSD 12a 0.60HPLLMRLLHHPSPSD 11a 0.70HPLLMRLLHHPS 13a 0.84

AcRHPILHRLLQPEG 17a 0.72AcRHPLHRLLQEAPS 18a 0.59RHPKILHRLLQPE 19a 0.62AcPILHRLLQEP 20a 0.69AcHPILHRLLQEP 21a 0.95AcHPLLMRLLLSP 22a 0.86CD spectroscopy measurements were performed under a constant nitrogen flow at 20 °C using 50 M of the peptide in

phosphate buffer (b) or in additional 30% (v/v) TFE (a). Information about helical propensity in semi quantitative manner couldbe shown by measuring the ratio of the characteristic wavelengths at 208 nm and 222 nm: 222nm 2208nm (c)

right: CD data for peptides 1, 4, 5, 13 and 22 (Table 4 2)

This view was reinforced by data from molecular modeling (MD) studies, which providedthe preferred folding of the peptides (Figure 4 5). In this case, the role of the flanking prolinesas helix breaker, and by that their ability to determine the precise helix length, was clearlyidentified. Introduction of prolines in the natural binder 1, for instance, resulted in areduction in the length of helical structure to the distance between the proline residues(Figure 4 5B). Furthermore, they predicted more influence concerning helicity in the regionof the LXXLL motif upon N’ terminal truncation in the case of 4 (resulting in 11) compared to10 (resulting in 12). Moreover, these MD simulations provide strong evidence that prolinesapply significant torsional limitations on the peptide backbone (most notably proline 2).

Chapter Four

80

Figure 4 5 | Structural analysis of proline derived peptide binders by MD simulation(A) Lowest energy conformation of peptides after 20 ns MD simulation (B D) helicity vs. peptidesequence derived from a 20 ns molecular dynamics (MD) simulation: (B) secondary structure vs.time plot of proline peptides for 1 (left) vs. 10 (right): grey ( helix), dark grey (310 helix), white(unordered). (C) comparison of peptides 10, 12 and 13; (D) comparison of peptides 4, 5 and 11.The degree of helical content per residue was obtained using the ptraj module of AMBER. Moredetails in 4.7 Experimental,Molecular Dynamics Simulation and Figure 4 11.

To provide further structural insight into the relative significance of the flanking prolineswithin the complex, the X ray structures of peptides 5 and 13 co crystallized with the LBD ofER in the presence of the natural ligand 17 estradiol (E2) were solved (Figure 4 6, 4 9 and4 12). In the case of peptide 13, both prolines (located at 2 and +3 flanking positions) aresituated above the charge clamp residues glutamic acid 493 (Glu493) and lysine 314 (Lys314)(Figure 4 6), respectively.

B

C D

A


81

Figure 4 6 | Co crystal structure of 13 ER LBD complex(A) LBD of ER with natural ligand estradiol (E2), helix 12 (H12, dark) of the activation function 2(AF 2), and 13, (B) zoom in on 13 (P 2, P+3) binding to AF 2 with charge clamp, including theglutamic acid (E493) and lysine (K314) residues.

An overlay of the structures for 13 and 1 (PDB 3OLS) both harboring a histidine residue atthe 3 position (His 3) revealed that for 13, the proline residue at position 2 (Pro 2) assistsin directing the histidine side chain residue toward glutamic acid 493 (Glu493) at ER surface,thereby enabling an additional stabilizing hydrogen bonding interaction (next toGlu332 His 3). Similar structural insights are observed in the corresponding co crystalstructure of 5, which contains Pro 2. Moreover, atomistically detailed dynamics simulationsshowed these additional hydrogen bonds are extraordinary stable, along the 6 ns simulation.This suggests that the proline with the histidine in front (HPX motif) might be stabilizing andmight lead therefore to higher affinity binding. Indeed, scission of the histidine from 21 to 20,resulted in an approximate 3 fold reduction in binding affinity (Table 4 2, Ki = 262 nM).Additionally, cellular data highlighted the importance of the prolines resulting in clearenhanced binding in the case of ER (Figure 4 7a). Mutational studies identified the prolinemediated hydrogen bond between the prior histidine and the glutamic acid (ER : E542; ER :E493) in the charge clamp being essential, since charge clamp mutation to alanine leads toloss of activity (Figure 4 7b).

ER -LBD E2

RD-29-66

P-2P+3

K314 E493

13ER AF-2 & 13

H12

A B

Chapter Four

82

Figure 4 7 | Influence of the prolines on ER binding in the M2H assayMammalian two hybrid studies on peptides 1, 21 and 22 comparing normalized luciferaseactivity (mean ± s.e.m.) for ER vs. ER . The SRC1 Box2 mutants with one proline residue ateither the 2 (1*) or +3 positions (1**) were also included: (A) with and without 17 estradiol (E2),and (B) wild type (wt) vs. ER charge clamp mutants E378A and E542A vs. AF 2 mutants E332Aand E493A. (n=3)

The biochemical, molecular dynamics simulations and structural data all combine tosuggest that peptides conforming to a PXLXXLLXXP consensus bind to ER more potentlythan conventional natural sequences. In this case, the prolines have evolved to function intwo ways: firstly as a ‘molecular ruler’, which primes the helix length to enable an optimal fitbetween the two charge clamp residues, Glu493 and Lys314, and secondly to nucleate andstabilize helix formation through additional conformational constraints on adjacent aminoacids.41

4.5. Discussion

Functional LXXLL motifs are not exclusive to NR coactivators, since they were alsoidentified in several transcription factors, including the calcium response element bindingprotein (CBP), the acetyltransferase p300, and other mediator subunits.42 Indeed, naturemakes frequent use of leucine rich helical structures to perform a variety of functions, suchas the HIV 1 accessory protein Vpr,43 which is important for viral production, and the wellcharacterized leucine zipper proteins such as c jun/c fos, which are necessary components oftranscription.44 By contrast, the helix averse proline is frequently located at the termini ofhelices in water soluble proteins,45 while in contrast, helical transmembrane proteins

such as GPCRs frequently make use of prolines to introduce kinks or perturbations into longhelical structures, which engender functionality through increased conformationalflexibility.46 According to Zimm Bragg/Lifson Roig helix coil transition theory, whereasleucine functions both to nucleate helix formation and propagate helix length, proline incontrast serves as helix nucleator and terminator.47,48 Furthermore, in proteins, proline isthought to form capping motifs at the N’ and C’ terminus of helices, respectively,49,50

A B


83

which contributes towards helix stability and the specificity of protein folding through thecorrect positioning of adjacent side chain residues.51 In this present work, cluster analysis ofthe ribosome hit sequences, however, could not identify any such known proline motifs;most likely due to the dependence of helix formation on NR surface binding. Indeed, thepredominance for lipophilic amino acid residues C’ terminal to Pro 2 is dominated by thelipophilic nature of the NR box. Residues N’ terminal to this Pro 2, however, do not showany characteristic enrichment. The important role of the charge clamp in this unique case canbe observed in the circular dichroism (CD) data, where the proline derived peptidesmeasured strong helicity only under conditions that mimic the hydrophobic environment ofthe charge clamp interaction (Table 4 2). The molecular modeling parameters used for thiswork have also been reported to mimic the effect of protein proximity on the peptidesecondary structure.52 Under such conditions, the peptides showed strong helical character,and the prolines were consistently located at the 2 and +3, or corresponding N1 and C’position (Figure 4 8).

Figure 4 8 | Designation of positions relative to helix or LXXLL motif.The top row shows labeling of amino acid positions with respect to the helix formation.Positions N1 to C1 are part of the helix, while N/C’, N/Ccap are in the flanking regions.53 The secondrow shows labeling of the amino acid position with respect to the consensus sequence.

Interestingly, position specific analysis of helices in globular proteins show prolines to behighly favored at position N1 and C’, respectively.53 Although not conforming to anypreviously identified motifs, the prolines are nonetheless postulated to nucleate helixformation and to stabilize the helix. Evidence for this can be found by careful examination ofthe crystallography data, where the precise anchoring of the prolines at their respectivepositions is thought to be critical in this case. The conformational constraints imposed by theprolines on adjacent amino acids results in a reorientation of side chain residues andbackbone carbonyl groups leading to more favorable interactions. At the N’ terminus thisresults in stabilization through additional interactions at the charge clamp, which can beevidenced by overlaying the co crystal structures of 5 and 1 bound to ER (Figure 4 9). Bothsequences possess a His 3 adjacent to the N’ terminal proline. In both peptides, the histidineresidue is seen making an H bonding interaction with Glu332. In the case of peptide 5,however, the presence of the proline in the peptide chain causes movement in the histidineside chain toward the NR charge clamp resulting in an additional stabilizing H bondinteraction with Glu493. In a number of cases, this additional H bonding interaction translatesinto a fine tuning of the binding in favor of greater affinity according to the fluorescencepolarization data. For example, insertion of prolines into 1, and subsequent truncation (21)

Chapter Four

84

leads to a statistically significant improvement in binding affinity while removal of theproline residues from 21 (peptide 24) or removal of the histidine residue (peptide 20) bothlead to a loss of activity. Cellular data confirm these findings through enhanced activity uponintroduction of Pro 2 into 1 (21 or 1*). Furthermore, the importance of the hydrogen bondbetween the prior histidine and the charge clamp glutamic acid (ER : E542; ER : E493) couldbe confirmed due to loss of activity upon E A mutation. Interestingly, the activity of 1against ER was reduced by only 50%, while 1 on ER and Pro 2 based peptides on bothsubtypes became completely inactive (Figure 4 7b). On the other hand, the E A mutationof the rear side glutamic acid residue (ER : E378; ER : E332) at the binding interface resultedin enhanced binding for Pro 2 based peptides, emphasizing the importance of the additionalor alternative hydrogen bond with the charge clamp. Surprisingly, the peptide 23, arationally designed peptide based on the most abundantly occurring amino acid sequencefrom cluster analysis of the ribosome display data, also featured significantly increasedligand independent activity (Figure 4 7a).

Figure 4 9 | Influence of the prolines on the helix ER interaction(A) Superimposition of histidine at the 3 position (His 3) of 5 and 1 (PDB, 3OLL). Prolineoriginated rearrangement of the prior His 3 enables formation of additional hydrogen bondbetween the protonated N1 and the charge clamp (E493) of ER . (B) The co crystal structurehighlights the role of the proline at the +3 position (Pro+3) in terms of its ability to cause a ‘kink’in the downstream amino acid sequence.

The role of prolines as helix initiator has also been identified in nature. Position specificanalysis on a set of proteins identified a X Pro motif (with X often a His) at the beginning ofhelices, in which the proline at the N1 position facilitates the formation of hydrogen bonds

involving the side chains of prior residues (Ncap).53 This directing role is comparable to thePro 2 found in the most active sequence identified by ribosome display screening, with thedifference being that the additional stabilizing hydrogen bond is formed intermolecular withthe surface charge clamp and not within the peptide helix (Figure 4 9).Additionally, this theoretical and experimental data is supported by sequence analysis of

NR coactivator proteins, where prolines are rarely found within the LXXLL motif, but morefrequently in the flanking regions. For instance, the two NR boxes of TRAP220/MED1 bothcontain Pro 2 (NPILTSLLQ) or even HPX (HPMLMNLLK),54,55 which is also the case for

His-3-5-

-1- His-3

K314 E493 E332

Pro+3 (-13-)

N1 N3

N1

A B


85

NF B inhibitor beta (NPILARLLR)56 and NCoA 6/ASC 2 (SPLLVNLLQ).57 The proclivity ofNRs to bind proline flanking peptide sequences is reflected in data from early phage displayscreening. Initially, the LXXLL coactivator motif was thought to function solely as a dockingmodule.58 Later, phage screening of the flanking regions of the LXXLL motif against ERindentified three distinct extended recognition motifs, either dependent or independent ofthe surface charge clamp. Although PXLXXLL was one of those three peptide classesidentified, only four analogues were cited in this case (one of which showed interestingselectivity for ER over ER ).26 In a separate piece of work, large diverse recombinantpeptide libraries were simultaneously screened against ER and TR . As for ER in theprevious example, only a few of the ER specific hit peptides (three of nineteen) harbored aPro 2. Somewhat intriguingly, however, is the fact that most of the peptide analoguesreported for TR (16 of 19) feature a PXLXXLL consensus, which is perhaps unsurprisinggiven its natural propensity to bind TRAP220.27,59 These reports, combined with the datapresented in this work hints at an evolutionary role played by proline residues for the finetuning of coactivator binding through control over helix length.Despite the enlightening sequence homology data and the groundbreaking phage display

work, there has been a lack of structural and biochemical evidence – especially co crystalstructure data – to explain the binding preference of proline flanking coactivator sequences atNR surfaces. Based on the evidence presented in this work, it could be speculated that thetendency for nature to introduce proline residues in the flanking regions of the LXXLLcoactivator motif may function as a fine tuning of binding properties. The resultingimproved selectivity or binding affinity might even override other determining factors, bycontrolling the length of the helix and increasing helix stability. The fact that thePXLXXLLXXP motif emerges early on during the ribosome display screening and is morepronounced based on the final cluster analysis for ER compared to ER (Figure 4 2)supports the hypothesis that receptor preference is a function of the proline residue’soccurrence int eh peptide sequence. Investigations of the binding affinity are in line with this,given the fact that the best hits indentified for ER overpower the SRC1 Box2 interaction forER , but not for ER (Table 4 1 and Figure 4 10). However, other factors such as charge andhydrophobicity are also certain to play a role in this protein protein interaction. For drugdiscovery, the tactical employment of flanking prolines may therefore have importantconsequences for the future development of potent and selective peptide based inhibitors ofthe NR coactivator interaction.

Chapter Four

86

Figure 4 10 | Comparison of Ki values of the proline peptidesComparison of Ki values (mean ± s.e.m.) for ER and ER determined in a competitivefluorescence polarization assay (values in Table 4 1). The dashed lines and arrows (dark =enhanced binding; light = decreased binding) facilitate comparison of peptide activities with thereference peptide 1.


87

4.6. Conclusion

In conclusion, a novel use of ribosome display has been described which effectively screensthe ER surface for novel peptide inhibitors. In this way, a series of hit peptide sequences werediscovered conforming to an advanced PXLXXLLXXP consensus motif, which was mostpronounced for ER . A selection of the hit sequences were prepared as protein fusions andas 19 mer peptides by SPPS and their binding to both ER and ER determined by an initialcellular screen (SNAP tag competition). According to circular dichroism measurements,while not intrinsically helical in pure phosphate buffer, the proline derived peptidesacquired significant helicity in 30% aq. TFE (v/v), suggesting that binding to the ER surface isa prerequisite for helix nucleation and stabilization. In this case, molecular modeling datahighlighted the role of the prolines in determining the precise helix length, which in theoptimal case is believed to match the distance measured between the two surface chargeclamp residues Lys314 and Glu493. Furthermore, X ray co crystallography data has clearlydemonstrated that the proline residues located at positions 2 and +3 (present in the mostpotent peptide binders) sit precisely above the surface charge clamp at the N1 and C’positions of the helix. Here, aside from their role as helix breakers, the conformationalconstraints imposed by the flanking prolines are postulated to restrict the orientation ofH bonding, carbonyl groups and adjacent amino acid residues, resulting in the improvedstabilization of the helix, as well as more favorable interactions at the ER surface. Thesefactors combined offer a structural rationale for the improved binding properties of theproline derived peptides (compared with the natural sequence, SRC1 Box2), which hasenabled the design of short (11 mer) and highly potent peptide inhibitors in thelow nM range. The initial higher occurrence of prolines in position 2 and +3 for ERcompared to ER is reflected in the ability to overpower SRC1 Box2 binding for ER , but notER . Therefore, the proline derived peptide inhibitors reported in this work represent a setof minimal structural parameters for addressing the ER coactivator interaction in preferentialmanner, based exclusively on natural amino acids and without recourse to additionalchemical modifications. Results from this work have thus laid important foundations for thefuture development of peptide derived tools and drug therapies with improvedpharmacological profiles.

Chapter Four

88

4.7. Experimental

GeneralRibosome display experiments and initial validation experiments were performed by Dr. Hoang D.Nguyen, Dr. Trang T.P. Phan and Monireh Goodarzifard . Peptide truncation studies were performedtogether with Dr. Matthew F. Burton, Stijn M. Agten and Dr. Lech Gustav Milroy. Molecular dynamicssimulations were conducted by Dr. Lidia Nieto. Solving the crystal structure was accomplished by Dr.Andrea Schmidt in cooperation with Dr. Rolf Rose and Dr. Christian Ottmann (MPI Dortmund,Germany). Cellular studies were performed by Dr. Ingrid J. de Vries van Leeuwen.

If not stated otherwise the following contents, descriptions, protocols, and distributer are valid:chemicals were ordered from SIGMA ALDRICH. Proteins were handled at 4 °C. Standard culturingmedium is lysogeny broth (LB) medium (10 g peptone, 10 g NaCl, 5 g yeast, 1 l water, autoclaved).SRC1 Box2 or short 1 is the 20 amino acid long sequence of the nuclear receptor coactivator 1(Ncoa1692 711) with the sequence SLTERHKILHRLLQEGSPSD, including a LXXLL motif.



construct vector tag protein originER LBD pET15b His6 hER LBD302 553 MSD, prior ORGANON

ER LBD pET15b His6 hER LBD260 502 BAYER HEALTHCARE PHARMACEUTICALS

AR LBD pGEX KG GST hAR LBD664 919 BAYER HEALTHCARE PHARMACEUTICALS

RXR LBD pGEX 4T1 GST hRXR LBD221 462 FOLKERSTSMA et al.60

Plasmids with the desired protein construct (Table 4 3) were transformed into E. coli BL21 cells.Emerging colonies were cultured over night in 25 ml LB medium with 100 g/ml ampicillin at 37 °C.This preculture was added to 2 l phosphate buffered, content rich terrific broth (TB) medium (12 gpeptone, 24 g yeast, 4 ml glycerol, 0.17 M KH2PO4, 0.72 M K2HPO4, 1 l water, autoclaved) withampicillin and incubated at 37 °C until an optical density at 600 nm (OD600) of 0.6 1.0. Addition of100 M Isopropyl D 1 thiogalactopyranoside (IPTG) induced expression of the protein of interest.For stabilization 10 M of ligand (ER, estradiol | AR, dihydrotestosteron | RXR, agonist LG) wasadded. After incubation for 20 h at 15 °C the cells were harvested by centrifugation (15 min at 7,000rpm) and the resulting pellet was stored at 80 °C until further usage.If not stated otherwise, the protein purification was performed at 4 °C. The pellet was resuspended inlysis buffer (LyB 1: 1x PBS tablet, CALBIOCHEM, 370 mM NaCl, 10% (v/v) glycerol, 0.1 mM TCEP,1 mM PMSF, 1 g/ml DNAseI, 10 M ligand, pH 8.0, His6 tag: additional 40 mM imidazole) and lyzedwith the Emulsi Flex C3 homogenizer (2 passes of 150,000 kPa, AVESTIN INC.). After anothercentrifugation step (40 min, 20,000 rpm) the supernatant was provided on the respective affinitychromatography column (GST tag: Protino GST/4B, 1 ml, MACHEREY NAGEL | His6 tag: His BindResin, 3 ml, NOVAGEN), washed with wash buffer (WaB 1: GST tag: 1x PBS tablet, CALBIOCHEM, pH 7.3| His6 tag: 1x PBS tablet, CALBIOCHEM, 370 mM NaCl, 40 mM imidazole, 10% (v/v) glycerol,0.1 mM TCEP, 10 M ligand, pH 8.0) and eluted with competing molecules (GST tag: 50 mM Tris,10 mM glutathione | His6 tag: WaB 1 plus 100 mM / 250 mM / 500 mM imidazole). Fractions wereanalyzed with SDS PAGE and optional with LC MS. The buffer of the combined fractions were


89

exchanged with Amicon Ultra 15 Centrifugal Filter Unit with Ultracel 10kDa membrane (MILLIPORE)to desalting buffer (20 mM Tris, 25 mM NaCl, 10% glycerol, pH 8.0) including 10 M ligand andlong term stored at 80°C.

Solid Phase Peptide Synthesis & PurificationThe synthesis of all peptides was performed on solid support, using the Fmoc strategy. If not statedotherwise, 200 mol (1 eq.) of Rink Amide MBHA resin (0.59 mmol/g, Novabiochem) was swollen for30 min in N methylpyrrolidone (NMP). The protective Fmoc group was removed by incubation with20% piperidine in NMP (2x 5min). Subsequently the amino acids (AA, 4 eq.) were activated using2 (1H Benzotriazole 1 yl) 1,1,3,3 Tetramethyluronium hexafluorophosphate (HBTU, 4 eq.), and N,NDiisopropylethylamine (DIPEA, 8 eq.). Single coupling was performed by shaking 45 min at roomtemperature. After finishing the desired peptide sequence, the resin was washed 3 times alternatelywith diethylether (DEE) and dichloromethane (DCM) and dried under vacuum for 10 min. The sidechain protection group and the cleavage of the peptide sequence from the resin were simultaneouslyachieved by the usage of mixture of trifluoroacetic acid, triisopropylsilane and water (TFA/TIS/H2O,95/2.5/2.5) for 3 h. The free peptide was precipitated by adding it drop wise to ice cold DEE. A finalcentrifugation step (2.000 rpm, 10 min) with subsequent washing (ice cold DEE), uptake in water andlyophilization results in the crude peptide. Purification was accomplished by reverse phase HPLC ona Alltima™ HP C18 column (125x20 mm, Alltech). Water (0.1% TFA) was used as a polar phase, addingdifferent amounts of an apolar acetonitrile (ACN, 0.1% TFA) phase. A linear gradient (20 ml/min) wasoptimized for each peptide to a 5% range, for instance 35 40% ACN. The purity of the peptide wasdeterminate at an analytical LC MS using GraceSmart RP18 (50x2.1mm, Grace, 3u, 120A) columnFinally, the peptides were lyophilized and stored at 80 °C.

Fluorescence Polarization and Depolarization assay61,62

Both fluorescence polarization (FP) and fluorescence depolarization (FDP) experiments weremeasured in black 384 well plates (PERKIN ELMER 384 F) on a Safire2 (TECAN) plate reader. The finalvolume of one well was 30 l containing constant concentrations of ligand (5 M) and Fluoresceinlabeled peptide (0.1 M, ER: NCoA 1/SRC1683 701, FL LTERHKILHRLLQEGSPSD; AR: FLETT 1,CSSRFESLFAGEKESR, phage display hit AR LBD;39 D22, LPYEGSLLLKLLRAPVEEV, INVITROGEN,phage display hit ER LBD, good binding properties RXR 26). In case of FP assay the protein wassequentially diluted (24 steps) in TR FRET coregulator buffer E (INVITROGEN). On the other hand,during FDP assays the protein concentration was kept constant (ER: 0.4 M; RXR, AR: 1 M), varyingthe concentration of the potential peptide based binder. Briefly, in 96 well plates dilution series wereprepared in 55 l and filled up with to 110 l with a 2X master mix solution containing all non varyingcomponents. Finally three times 30 l was transferred to 384 well plates, centrifuged (1,000 rpm,2 min) and incubated for 1 h at 4 °C. Plates were measured 50 times at 30 °C (excitation 470 nm,emission 519 nm). Polarization was plotted against the concentration of either the protein (FP) orthe test peptide (FDP), whereby each data point represents an average of 3 experiments.The dissociation constant Kd of the protein peptide complex (FP) was calculated using equation 1a,b:


Chapter Four

90

In the competitive setup (FDP) the half maximal inhibitory concentration ( ) was calculated withthe means of equation 1a and 2:

The inhibitory constant Ki of the different peptides could be determinated with the value of theprotein (from equation 2) and equation 3a,b.

where Kd is the dissociation constant of the fluorescein labeled peptide protein complex (equation 1),y0 is the initial bound to free concentration ratio for labeled peptide, [AB] is the concentration ofprotein peptide complex, [A] the concentration of unbound peptide, and [B] the concentration ofprotein (ER).

Ribosome DisplayIf not stated otherwise, protein expression and purification were conducted like described above in‘Protein Expression & Purification’. The following constructs were used to express the ER ( / ) LBDsuitable for surface immobilization.

Table 4 4 | Used construct for ribosome display

construct vector tag protein originpHT503 pTriEx 4 Strep His6 ER –LBD302 553 CARRAZ et al.63

pHT504 pTriEx 4 Strep His6 ER LBD260 502 NGUYEN et al.64

Where His6 tag was used for purification and Strep tag for immobilization. After induced proteinexpression in E. coli Rosetta 2 (DE3) pLacI cells (NOVAGEN), a 5 ml His Trap HP 5 column(GE HEALTHCARE) was used for purification. SDS PAGE and LC ESI MS analysis confirmed the ERLBD and ER LBD containing fractions. Finally the purified proteins were immobilized on StrepTactin coated plate (IBA) and the activity was determined by its SRC1 Box2 eCFP binding ability. TheDNA library was designed based on the known ER binder SRC1 Box2:

gcc atg aat tct ctg acc gca cgcNNK NNK NNK NNK NNK cgt NNK NNK NNK NNK NNK

agc ccg tct gat aag ctt gcg gccWhere N is any base and K is guanine (G) or thymidine (T). The plasmid with the randomized region,resulting in SLTARXXXXXRXXXXXSPSD peptides (with X, any amino acid), was expressed in vitrousing PureExpress In Vitro Protein Synthesis kit (NEB). In a 96 well plate format the library ofribosomal complexes, consisting of expressed peptide tethered to ribosome with unreleased mRNA,was screened against immobilized ER LBD and ER LBD, respectively. The narrowed library withconnected geno and phenotype was used as basis for the following reversed transcription resulting inoptimized DNA library with enhanced ER binding properties. After several of these selection roundsthe library was processed and analyzed as followed: introduction in vector using In Fusion TMAdvantage PCR Cloning kit (CLONTECH), transformation in E. coli OmniMax, confirmation by colonyPCR, sequencing, conversion to analogue amino acids and alignment by means of Weblogo.65,38


91

Selected hits:2, LTARSSLLIRLLQLPSPSD | 3, LTARMPQLTRLLLSPSPSN4, LTARSPLLTRLLLQPSPSD | 5, LTARHPLLLRHLLQNSPSD6, LTARLLMSSRLITLLSPSD | 10, LTARHPLLMRLLHHPSPSD

Circular Dichroism spectroscopy 66

Far UV Circular Dichroism (CD) spectroscopy measurements were performed under a constantnitrogen flow at 20 °C using 50 M of the peptide in the presence of 30% (v/v) trifluoroethanol (TFE),if not stated otherwise. As a baseline a peptide free buffer (30% TFE) was measured under identicalconditions. The spectrum was recorded from 250 nm to 185 nm using a JASCO 815 spectrometer.Quartz cuvettes with path lengths of 1 mm or 0.2 mm (HELLMA ANALYTICS) were employed with adata pitch of 0.5 in a continuous mode, a scan speed of 20 nm/min with a response time of 2 s and abandwidth of 0.5 nm. The graphs are representing an average of five scans. The observed ellipticity(degrees, in [mdeg]) was converted to mean residue ellipticity (in [deg cm2/dmol]) usingequation 3.

Where is the molecular mass of the peptide (in [Da]), is the number of amino acids in the peptide,is the path length (in [cm]), and is the peptide concentration (in [g/ml]). Information about helical

propensity in semi quantitative manner could be shown by the ratio of the characteristic wavelengthsat 208 nm and 222 nm: .

Molecular Dynamics SimulationAll MD simulations were carried out using the AMBER suite of programs, and the ff03 force field.67 Animplicit solvent was used via the General Born solvation method (IGB 5), as implemented in AMBER.68

Starting with an extended initial conformation, built by the LEaP module of AMBER for each peptide,all MD simulations were fully unrestrained and all trajectories were generated using the sanderprogram in the AMBER 9 package. Each peptide was simulated for a total of 20 ns at 51.9 °C, followinga similar approach as previously reported to predict the conformation of a miniprotein.69 Analysis wasperformed on the 20 ns using 20,000 trajectory snapshots spaced every 1 ps. A snapshot with thelowest potential energy across the simulation was chosen as representative structure for each peptide.The secondary structure of each residue as a function of time was subsequently analyzed utilizing theSTRIDE secondary structure assignment algorithms as implemented in VMD.70,71 All MD simulationswere performed by Dr. Lidia Nieto.

Chapter Four

92

Figure 4 11 | Degree of helical content per residue and structure based on MD simulationStarting from an extended initial conformation, MD simulations were fully unrestrained and alltrajectories were generated. Each peptide was simulated for a total of 20 ns 20,000 trajectorysnapshots spaced every 1 ps at 51.9 °C (325 K). (A, B) Grayscale is representing the propensityforming secondary structures: white, unordered; light grey, helix; dark grey, turn; black, 310 helix.Dashed line shows start/end of helix. A snapshot with the lowest potential energy across thesimulation was chosen as representative structure for each peptide: (A) Natural binder SRC Box2(1), (B) after introduction of prolines in 2 and +3 position (1 Pro). (C) The degree of helicalcontent per residue (for 1 and 1 Pro) was obtained using the ptraj module of AMBER omitting thefirst 5 ns of each simulation. (D) Predicted structure of peptides 5 and 13 based on MDsimulation.

Co crystallizationIf not stated otherwise, protein expression and purification were conducted like described above in‘Protein Expression & Purification’.

Table 4 5 | Used construct for co crystallization

construct vector tag protein originER LBD pET16b ER LBDMD(D261 L500)DD MÖCKLINGHOFF et al.72

A 1 B 1-Pro

C D

5 1311-Pro


93

After induced protein expression in E. coli cells, they were harvested, lysed (LyB 2: 20 mM Tris, 0.5 MNaCl, 5 mM DTT, 1 mM EDTA, pH 7.5), and subsequently the tag free protein was purified with a3 ml estradiol sepharose column (PTI RESEARCH, INC.). After washing (250 ml WaB 2: 10 mM Tris,0.5 M NaCl, 5 mM DTT, 1 mM EDTA, pH 7.5) cysteins of the protein were S carboxymethylated using5 mM iodoacetic acid (50 ml WaB 2, but 0.2 M NaCl, 1 h, 4 °C). After an additional washing step(400 ml WaB 2) the pure protein was eluted with step wise increasing amounts of estradiol (WaB 2,100 M, 150 M, 200 M E2) and by means of a centrifugal 10 kDa cut off filter unit (AMICON) bufferexchanged (WaB 2, but 0.2 M NaCl, 5 mM DTT) and concentrated to 8 12 mg/ml. Characterizationwas accomplished by SDS PAGE and photometric determination of the concentration (NANODROP,280 nm).11 mg/ml (400 M) of the protein was mixed with the peptide (5 or 13) in a 1 : 1.5 molecular ratio (600M) and incubated for 30 min at 4 °C. In order to crystallize this peptide protein complex, it was

combined with the respective well solution in a 2 : 1 ratio. Finally, a total volume of 4.5 l wasequilibrated at 4 °C against 1 ml reservoir solution on a 24 well crystallization plate (NEXTAL) usingthe hanging drop crystallization methodology. For initial screens the JCSG+ suite (QIAGEN), whichconsists of different premixed crystallization conditions, was used. Further optimization wasperformed with the suites JCSG Core I, Core II, Core II, and Core IV. Optimal conditions for crystalgrowing were 10% PEG6000, 1 M LiCL, 100 mM citric acid, and 20% glycerol at pH 5. For X rayanalysis single crystals were transferred into a CryoLoop (HAMPTON RESEARCH) and directly cryocooled in liquid nitrogen, since the optimal condition already consists of 20% glycerol.In order to solve the crystal structure, molecular replacement of a ER LBD/ligand/peptide complex(PDB code: 1U9E73) using the program PHASER was performed.74 On a Nonius AXS MICRO A (MARdtb detector, crystal to detector distance was 150 mm, oscillation range was 1.0°) a native data setfrom an optimized ER LBD peptide complex crystal was collected at 173 °C (100 K). After dataindexing and integrating, the XDS package was used for scaling. The crystals diffracted to a maximalresolution of 2.1 Å (5) and 2.3 Å (13), respectively. Furthermore the crystals were assigned to spacegroup C121 (5) and P3(2) (13) with 4 (5) and 2 (13) monomers per asymmetric unit. Statistics of bothcrystal structures are listed in TAB.

Chapter Four

94

Table 4 6 | X ray conditions

Peptide 5 13general settings

Protein ER LBDMD(D261 L500)DD

Ligand estradiolBeamline / generator Nonius / Bruker AXS MICRO StarWavelength [Å] 15.418Temperature [K] 173 °C (100 K)

peptide specific settingsAmino acid sequence LTARHPLLLRHLLQNSPSD HPLLMRLLHHPSNo. of residues 19 12No. of crystals with protein 1 1Space group C 121 P3(2)

Unit cell dimensions

a = 192.98 Å a = 62.63 Åb = 53.49 Å b = 62.63 Åc = 131.39 Å c = 125.67 Å

= = 90°, = 120.47° = = 90°, = 120.47°Resolution range / (outer shell) [Å] 20 2.1 (2.3 2.1) 20 2.3 (2.6 2.3)No. of observations 246056 (57347) 45679 (14205)No. of unique reflections 67629 (15989) 23805 (7512)Redundancy overall / (outer shell) 3.64 (3.59) 1.91 (1.89)Completeness overall / (outer shell) [%] 99.2 (99.1) 97.2 (99.4)<I/ (I)> overall / (outer shell) 15.57 (5.17) 15.32 (5.38)Rmerge1 overall / (outer shell) 10.7 (35.4) 8.3 (27.2)Maximal Resolution 2.10 Å 2.30 ÅNo. of reflections 64245 22612Rcryst2 / Rfree3 20.48 / 24.98 21.00 / 26.46

No. of atoms / residuesMonomer per asymmetric unit 4 2Protein 7265 / 899 3624 / 456Peptide 388 / 44 197 / 22Others 80 / 4 40 / 2Water 416 154Average B factors2 31.98 Å 35.82 Å

rmsd from ideal geometryBond lengths 0.022 Å 0.018 ÅBond angles 1.80° 1.82°

Ramachandran plot statistics (%)Most favored regions 95.6 96.1Addionally allowed regions 3.6 3.5Generously allowed regions 0.4 0.5Disallowed regions 0.1 0

1Rmerge= hkl j|Ihkl(j) (Ihkl)|/ hkl jIhkl(j) ,where Ihkl is the mean density and Ihkl(j) are individual intensity measurements of thereflection hkl 2Rcryst = hkl |Fhkl(obs) Fhkl(calc)|/ hklFhkl(obs) 3Rfree is the same as R factor Rcryst but calculated with 5% of the

reflections that were excluded from refinement.

The program Coot was used to calculate the electron density maps.75 In order to perform the iterativestructure refinement the program REFMAC (CCP4 suit) was used.76


95

Figure 4 12 | Stereo image of a portion of the electron density mapER LBD in complex with peptide and peptide close up (A,B, 5; C,D, 13)

A

B

C

D

Chapter Four

96

Mammalian Two Hybrid assayAll Mammalian Two Hybrid assays (M2H) were performed in human osteosarcoma cells (U2 OS77)seeding 30,000 cells/well in a 24 well plate (BD Biosciences). Plasmids are either from MammalianTwo Hybrid Assay Kit (AGILENT) or cloned by Dr. Ingrid J. de Vries. Culturing conditions were 37 °Cand 5% CO2. Cells were transfected with following amounts of DNA/well using linearpolyethylenimines (PEI, POLYSCIENCES): (1) 40 ng of the bait construct pCMV BD PEP containingpeptide of interest (PEP) fused to the DNA binding domain of the GAL4, (2) 40 ng of the targetconstruct pCMV AD ER / , containing ER ( or ) fused to the transcriptional activation domain(AD) of the mouse protein NF B, (3) 200 ng of the reporter plasmid pFR Luc with the syntheticpromoter with five tandem repeats of the GAL4 binding sites that control expression of the Photinuspyralis firefly luciferase, and (4) 3.2 ng of control plasmid Renilla Luc, expressing independently Renillareniformis luciferase. After 16 h incubation cells were treated with 10 nM estradiol and after additional24 h incubation the interaction was determined with a Dual Luciferase Reporter Assay (PROMEGA),according to manufacturer’s instruction. The luminescent intensities were recorded on a Synergy HTMulti Mode Microplate Reader (BIOTEK). The interaction dependent Firefly luciferase signal was firstnormalized over the independent Renilla luciferase signal and subsequently normalized over the ratioof untreated cells. Studies were performed by Dr. Ingrid J. de Vries van Leeuwen.Displacement assay (SNAP tag) in cells were also based on M2H. Co transfection of pCMV AD ER /and pCMV BD 1 produce normal signal of luciferase representing for the binding of SRC1 Box2 to ERLBD. Titration of a SNAP PEP plasmid carrying the sequence of the PEP fused to a SNAP tag and anuclear localization signal – and the consequent decrease of luciferase signal is a function of the abilityof the PEP to compete with SRC1 Box2 containing target construct (pCMV BD 1). Studies wereperformed by Dr. Hoang D. Nguyen.


97

4.8. References

1. C. L. Smith and B. W. O’Malley, Endocr Rev, 2004, 25, 45–71.2. I. P. Uray and P. H. Brown, Expert Opin Investig Drugs, 2006, 15, 1583–1600.3. V. C. Jordan, Br J Pharmacol, 2006, 147, S269–S276.4. V. C. Jordan, Br J Pharmacol, 1993, 110, 507–517.5. J. Millour, D. Constantinidou, A. V. Stavropoulou, M. S. C. Wilson, S. S. Myatt, J. M. M.

Kwok, K. Sivanandan, R. C. Coombes, R. H. Medema, J. Hartman, A. E. Lykkesfeldt, andE. W. F. Lam, Oncogene, 2010, 29, 2983–2995.

6. T. W. Moore, C. G. Mayne, and J. A. Katzenellenbogen,Mol Endocrinol, 2010, 24, 683–695.7. T. W. Moore and J. A. Katzenellenbogen, in Annual Reports in Medicinal Chemistry,

Academic Press, 2009, vol. Volume 44, pp. 443–457.8. L. Nagy and J. W. R. Schwabe, Trends Biochem Sci, 2004, 29, 317–324.9. D. M. Heery, E. Kalkhoven, S. Hoare, and M. G. Parker, Nature, 1997, 387, 733–736.10. Y. Li, M. H. Lambert, and H. E. Xu, Structure, 2003, 11, 741–746.11. J. M. Hall, C. Y. Chang, and D. P. McDonnell,Mol Endocrinol, 2000, 14, 2010–2023.12. P. Vlieghe, V. Lisowski, J. Martinez, and M. Khrestchatisky, Drug Discov Today, 2010, 15.13. H. B. Zhou, M. L. Collins, J. R. Gunther, J. S. Comninos, and J. A. Katzenellenbogen,

Bioorg Med Chem Lett, 2007, 17, 4118–4122.14. A. L. Rodriguez, A. Tamrazi, M. L. Collins, and J. A. Katzenellenbogen, J Med Chem, 2004,

47, 600–611.15. A. B. Williams, P. T. Weiser, R. N. Hanson, J. R. Gunther, and J. A. Katzenellenbogen, Org

Lett, 2009, 11, 5370–5373.16. A. Sun, T. W. Moore, J. R. Gunther, M. S. Kim, E. Rhoden, Y. Du, H. Fu, J. P. Snyder, and

J. A. Katzenellenbogen, ChemMedChem, 2011, 6, 654–666.17. E. Estébanez Perpiñá, L. A. Arnold, A. A. Arnold, P. Nguyen, E. D. Rodrigues, E. Mar, R.

Bateman, P. Pallai, K. M. Shokat, J. D. Baxter, R. K. Guy, P. Webb, and R. J. Fletterick, ProcNatl Acad Sci U S A, 2007, 104, 16074–16079.

18. A. L. LaFrate, J. R. Gunther, K. E. Carlson, and J. A. Katzenellenbogen, Bioorg Med Chem,2008, 16, 10075–10084.

19. B. Vaz, S. Möcklinghoff, S. Folkertsma, S. Lusher, J. de Vlieg, and L. Brunsveld, ChemCommun (Camb), 2009, 5377–5379.

20. T. Phan, H. D. Nguyen, H. Göksel, S. Möcklinghoff, and L. Brunsveld, Chem Commun(Camb), 2010, 46, 8207–8209.

21. A. K. Galande, K. S. Bramlett, J. O. Trent, T. P. Burris, J. L. Wittliff, and A. F. Spatola,Chembiochem, 2005, 6, 1991–1998.

22. M. C. Via, Insight Pharma Reports, 2010.23. C. Bracken, J. Gulyas, J. W. Taylor, and J. Baum, J Am Chem Soc, 1994, 116, 6431–6432.24. A. Patgiri, K. K. Yadav, P. S. Arora, and D. Bar Sagi, Nat Chem Biol, 2011, 7, 585–587.25. G. P. Smith, Science, 1985, 228, 1315–1317.26. C. y Chang, J. D. Norris, H. Grøn, L. A. Paige, P. T. Hamilton, D. J. Kenan, D. Fowlkes,

and D. P. McDonnell,Mol Cell Biol, 1999, 19, 8226–8239.27. J. P. Northrop, D. Nguyen, S. Piplani, S. E. Olivan, S. T. S. Kwan, N. F. Go, C. P. Hart, and

P. J. Schatz,Mol Endocrinol, 2000, 14, 605–622.

Chapter Four

98

28. L. C. Mattheakis, R. R. Bhatt, and W. J. Dower, Proc Natl Acad Sci U S A, 1994, 91, 9022–9026.

29. J. Hanes and A. Plückthun, PNAS, 1997, 94, 4937–4942.30. S. Fields and O. Song, Nature, 1989, 340, 245–246.31. M. G. Cull, J. F. Miller, and P. J. Schatz, Proc Natl Acad Sci U S A, 1992, 89, 1865–1869.32. E. T. Boder and K. D. Wittrup, Nat Biotechnol, 1997, 15, 553–557.33. J. A. Francisco, C. F. Earhart, and G. Georgiou, Proc Natl Acad Sci U S A, 1992, 89, 2713–

2717.34. C. Zahnd, P. Amstutz, and A. Plückthun, Nat Methods, 2007, 4, 269–279.35. M. He and M. J. Taussig, Nat Methods, 2007, 4, 281–288.36. B. Dreier and A. Plückthun, in Ribosome Display and Related Technologies : Methods and

Protocols, Springer, 2012, pp. 261–286.37. L. M. Yang, J. L. Wang, L. Kang, S. Gao, Y. Liu, and T. M. Hu, PLoS ONE, 2008, 3, e2092.38. G. E. Crooks, G. Hon, J. M. Chandonia, and S. E. Brenner, Genome Res, 2004, 14, 1188–

1190.39. E. Hur, S. J. Pfaff, E. S. Payne, H. Grøn, B. M. Buehrer, and R. J. Fletterick, PLoS Biol, 2004,

2, e274.40. P. Luo and R. L. Baldwin, Biochemistry, 1997, 36, 8413–8421.41. A. J. Doig, in Protein Folding, Misfolding and Aggregation: Classical Themes and Novel

Approaches Protein Folding, Misfolding and Aggregation: Classical Themes and NovelApproaches, Royal Society of Chemistry, Cambridge, 2008, pp. 1–27.

42. M. J. Plevin, M. M. Mills, and M. Ikura, Trends Biochem Sci, 2005, 30, 66–69.43. M. P. Sherman, C. M. de Noronha, D. Pearce, and W. C. Greene, J Virol, 2000, 74, 8159–

8165.44. W. H. Landschulz, P. F. Johnson, and S. L. McKnight, Science, 1988, 240, 1759–1764.45. M. W. MacArthur and J. M. Thornton, J Mol Biol, 1991, 218, 397–412.46. E. B. Van Arnam, H. A. Lester, and D. A. Dougherty, ACS Chem Biol, 2011, 6, 1063–1068.47. B. H. Zimm and J. K. Bragg, J Chem Phys, 1959, 31, 526–535.48. S. Lifson and A. Roig, J Chem Phys, 1961, 34, 1963–1974.49. A. R. Viguera and L. Serrano, Protein Sci, 1999, 8, 1733–1742.50. J. Prieto and L. Serrano, J Mol Biol, 1997, 274, 276–288.51. M. J. I. Andrews and A. B. Tabor, Tetrahedron, 1999, 55, 11711–11743.52. C. R. Gregor, E. Cerasoli, P. R. Tulip, M. G. Ryadnov, G. J. Martyna, and J. Crain, Phys

Chem Chem Phys, 2011, 13, 127–135.53. S. Kumar and M. Bansal, Proteins, 1998, 31, 460–476.54. C. Rachez, M. Gamble, C. P. Chang, G. B. Atkins, M. A. Lazar, and L. P. Freedman, Mol

Cell Biol, 2000, 20, 2718–2726.55. V. Pogenberg, J. F. Guichou, V. Vivat Hannah, S. Kammerer, E. Pérez, P. Germain, A. R.

de Lera, H. Gronemeyer, C. A. Royer, and W. Bourguet, J Biol Chem, 2005, 280, 1625–1633.56. J. W. Lee, H. S. Choi, J. Gyuris, R. Brent, and D. D. Moore, Mol Endocrinol, 1995, 9, 243–

254.57. J. L. Vanhooke, M. M. Benning, C. B. Bauer, J. W. Pike, and H. F. DeLuca, Biochemistry,

2004, 43, 4101–4110.58. E. M. McInerney, D. W. Rose, S. E. Flynn, S. Westin, T. M. Mullen, A. Krones, J.

Inostroza, J. Torchia, R. T. Nolte, N. Assa Munt, M. V. Milburn, C. K. Glass, and M. G.Rosenfeld, Genes Dev, 1998, 12, 3357–3368.


99

59. V. H. Coulthard, S. Matsuda, and D. M. Heery, J Biol Chem, 2003, 278, 10942–10951.60. S. Folkertsma, P. I. van Noort, A. de Heer, P. Carati, R. Brandt, A. Visser, G. Vriend, and

J. de Vlieg,Mol Endocrinol, 2007, 21, 30–48.61. P. Wu, M. Brasseur, and U. Schindler, Anal Biochem, 1997, 249, 29–36.62. F. Perrin, J Phys Radium, 1926, 7, 390–401.63. M. Carraz, W. Zwart, T. Phan, R. Michalides, and L. Brunsveld, Chem Biol, 2009, 16, 702–

711.64. H. D. Nguyen, T. T. P. Phan, M. Carraz, and L. Brunsveld,Mol Biosyst, 2012.65. T. D. Schneider and R. M. Stephens, Nucleic Acids Res, 1990, 18, 6097–6100.66. S. M. Kelly, T. J. Jess, and N. C. Price, Biochim Biophys Acta, 2005, 1751, 119–139.67. Y. Duan, C. Wu, S. Chowdhury, M. C. Lee, G. Xiong, W. Zhang, R. Yang, P. Cieplak, R.

Luo, T. Lee, J. Caldwell, J. Wang, and P. Kollman, J Comput Chem, 2003, 24, 1999–2012.68. A. Onufriev, D. Bashford, and D. A. Case, Proteins, 2004, 55, 383–394.69. C. Simmerling, B. Strockbine, and A. E. Roitberg, J Am Chem Soc, 2002, 124, 11258–11259.70. D. Frishman and P. Argos, Proteins, 1995, 23, 566–579.71. W. Humphrey, A. Dalke, and K. Schulten, J Mol Graph, 1996, 14, 33–38, 27–28.72. S. Möcklinghoff, R. Rose, M. Carraz, A. Visser, C. Ottmann, and L. Brunsveld, Chem Bio

Chem, 2010, 11, 2251–2254.73. E. S. Manas, R. J. Unwalla, Z. B. Xu, M. S. Malamas, C. P. Miller, H. A. Harris, C. Hsiao, T.

Akopian, W. T. Hum, K. Malakian, S. Wolfrom, A. Bapat, R. A. Bhat, M. L. Stahl, W. S.Somers, and J. C. Alvarez, J Am Chem Soc, 2004, 126, 15106–15119.

74. A. J. McCoy, R. W. Grosse Kunstleve, L. C. Storoni, and R. J. Read, Acta Crystallogr D BiolCrystallogr, 2005, 61, 458–464.

75. P. Emsley and K. Cowtan, Acta Crystallogr D Biol Crystallogr, 2004, 60, 2126–2132.76. G. N. Murshudov, A. A. Vagin, and E. J. Dodson, Acta Crystallogr D Biol Crystallogr, 1997,

53, 240–255.77. Y. Liang, D. F. Robinson, J. Dennig, G. Suske, and W. E. Fahl, J Biol Chem, 1996, 271,

11792–11797.

CHAPTER FIVE In Silico Design of Androgen Receptor Coactivator

Binding Inhibitors

Abstract. Molecular Dynamics (MD) Simulation was used to predict the propensity ofshort peptides to fold into an helix: one of the requirements for targeting the androgenreceptor (AR) coactivator binding surface. The rational in silico design of the peptides wasbased on separate results from estrogen receptor (ER) screening, which identified optimizedleucine rich motif with flanking prolines as helix primers (Chapter 4). The scaffold sequencewas based on a combination of the natural SRC1 Box2 sequence and the N’ terminal domain(NTD) of AR, later known to specifically interact with the AR ligand binding domain (LBD).Instead of a leucine rich sequence (LXXLL) – required for binding to ER and several othernuclear receptors (NRs) – a phenylalanine rich motif (FXXLF) was introduced to promote ARselective binding. Proline residues were inserted into the flanking regions of the centralbinding motif. Molecular dynamics (MD) simulation studies were performed to acquireinformation about the propensities of the designed peptides to form helices. Based on theseresults the solid phase synthesis of selected peptides was performed. Circular dichroismstudies of these peptides clearly confirmed the predicted helical character of the 11 and12 mers. The AR binding affinities of the peptides was evaluated by varying the length of thehelical segment of the peptides via the proline positioning and by elaborating the adjacentamino acid of the N’ terminal proline in terms of charge, size, and hydrophobicity. Thisconceptual approach resulted in short peptides that interact exclusively with AR LBD withKd/Ki values in the low M range, and that could even overpower known AR specificbinding in a cellular context with the fully intact AR. Thus, the identified sequences can beconsidered as lead structures for the future development of AR selective coactivator bindinginhibitors (CBIs).

Chapter Five

102

5.1. Introduction

The androgen receptor (AR, NR3C4) is part of the subclass of steroid receptors of thenuclear receptor (NR) superfamily, which also includes glucocorticoid , mineralocorticoid ,progesterone , and estrogen receptors.1 Misregulation of AR signaling has classically beentreated by either reducing androgens at the steroidogenesis level or through blocking of thebinding of the endogenous ligand, dihydrotestosterone (DHT), to the AR ligand bindingpocket (LBP).2 While initial steroid based anti androgens produced off target effects on otherNRs, the non steroidal small molecules hydroxyflutamide (Eulexin, SCHERING PLOUGH),bicalutamide (Casodex, ASTRAZENECA) and nilutamide (Nilandron, AVENTIS

PHARMACEUTICALS) are known to be selective androgen receptor modulators (SARMs).3,4 Thefirst co crystal structure of the AR bicalutamide complex provided an entry point for therational drug design of small molecules antagonists of increased potency and efficiency.5,6

The drawback of such compounds, however, is that long term treatment can result in the lossof antagonistic effect due to amino acid mutations in the AR Ligand binding domain (LBD),as well as reduced activity due to AR and AR coactivator up regulation or modification(post translational modification), or improvement of ligand (DHT) production.7 Thus,alternative to the classical approach targeting the AR ligand binding pocket (LBP), theprotein protein interaction (PPI) between the AR LBD and its coactivator – via activationfunction 2 (AF 2) is currently under investigation as a controllable and druggable interface.The human proteome consists of more than 375,000 PPIs.8 The emergence of PPIs as drug

targets for drug development has been aided by X ray crystallography, NMR and electronmicroscopy, which has enabled the atomistic level determination of large multiproteincomplexes. Additionally, next generation sequencing and high throughput genomicmapping has helped to identify peptide sequences in certain PPIs that are involved in certaindisease states.9 The use of biologics such as peptides for targeting PPIs has been validatedthrough the increasing numbers of approved drugs in recent years.10,11

The AR LBD exhibits only moderate sequence homology compared with the LBD of otherNRs (e.g. 50% compared to the mineralocorticoid (MR) and progesterone receptor (PR) oraround 20% in the case of the estrogen receptors subtypes and , ER / ).12 However,overall, they share a similar structure. Additionally, AR also possesses some uniqueattributes, especially at the interface with coactivator proteins. While the common ability ofNRs to bind leucine rich motifs (LXXLL) also includes AR, binding studies using the fulllength AR has revealed the extraordinary influence of the receptor’s N’ terminal domain(NTD). NR positive interaction partners show limited binding properties for AR, suggestingthat the activation function 2 (AF 2) of AR has dissimilar sequence specificity compared withother NRs.13,14 Detailed AF 2 investigations discovered, indeed, an ability for AR toaccommodate aromatic amino acid based sequences, including the FXXLF bindingpattern.15,16 This unique motif is also present in the AR NTD, which is reported to interact

In Silico Design of Androgen Receptor Coactivator Binding Inhibitors

103

with the AR LBD (AR N/C interaction). Furthermore, the charge clamp residues lysine 720and glutamic acid 897 of the AR AF 2 can interact with AR specific motifs forming twohydrogen bonds with their backbone, compared to only one hydrogen bond with LXXLLbased peptides.15 17 In a separate piece of work, the importance of the FXXLF motifs incoactivator recruitment was demonstrated, including members of the AR associated (ARA)coactivator family or the checkpoint protein hRAD9. The introduction of LXXLL motifs inthese coactivators resulted in the complete loss of AR binding ability.18,19 Vice versa, peptidesfeaturing FXXLF motifs show no binding to other NRs, except for PR LBD, whichnevertheless shows preferential binding to leucine based sequences.17 In general, structuralanalysis of the interface between NRs and their coactivators showed that the peptide exists asan helix of well defined length, which is determined by the charge clamp.15,20

In a previous study of the AF 2, the AR LBD was screened for novel peptide sequencesresulting in phenylalanine based motifs, but not the typical LXXLL NR motif.21,22 Thus, drugdiscovery is using this atypical surface pattern featuring the ability to accommodate bulkyaromatic residues to target specifically AR. Small molecule design through a mimic of thearomatic phenyl rings discovered CBIs with Ki values in the low M range.23–25 Other studieshave focused on peptide based coactivator binding inhibitors: for example the introductionof an FXXLF motif into helical miniproteins stabilized by intramolecular disulfidebonds.26,27 In general, stabilization of peptide secondary structures (e.g. bymacrolactamization,28 hydrocarbon stapling,29 or the incorporation of hydrogen bondingsurrogates,30) can lead to enhanced binding properties due to preorganization of the helixin a surface bound like state. Nonetheless, the drawbacks of such approaches are thenecessity for synthetic operations to upgrade amino acid building blocks to facilitatemacrocyclization and the unpredictable effects that such modifications might have onbinding.In the present study, potential AR CBIs were designed in silico, making recourse to both the

well characterized specificity of the unique FXXLF motif for AR AF 2 binding and theinfluence of prolines on helix formation and stability (PXLXXLLXXP) discovered forER AF 2 interaction (chapter 4). The required degree of helicity was initially monitored bymolecular dynamics (MD) studies and later confirmed by CD spectroscopy of peptides madeby solid phase peptide synthesis (SPPS). Finally, the potential CBIs were validated inpolarization based binding studies and using cell based assays.

Chapter Five

104

Figure 5 1 | Principle of rational in silico designRational Design by combining knowledge from two research fields: (top left) ER ribosomescreening (PXLXXLLXXP), investigations into the influence of the prolines on helix formationand stability; (top right) selective AR binding (FXXLF), the specificity of the unique phenylalaninerich consensus motif; (middle) structural comparison of AF 2s (in silico); (I.) propensity of helixformation by MD simulation; (II.) SPPS and subsequent structural analysis (CD) and bindingstudies (polarization assay, mammalian one hybrid).

5.2. In silico peptide design

In a prior study, the LBDs of ER and ER were screened for novel binding motifs usingthe ribosome display technique. The main advantage of ribosome display over otherscreening methods in particular phage display is the exceptionally high diversity of theDNA library, since the screening is performed entirely in vitro.31 Results from this proteinsurface screen resulted in a consensus motif featuring the hallmark LXXLL sequence –known to be important for NR binding – and additional prolines in the flanking regions ofthe core motif (chapter 4). Transferring these investigations to AR requires a detailedcomparison of both coactivator binding surfaces (AF 2) with respect to size, hydrophobicityand charge, since the total sequence homology between ER and AR is relatively low (18.6%)compared to other NRs.


105

Figure 5 2 | AF 2 of AR compared to ERShown are amino acids of the ligand binding domain involved in the generation of the AF 2, thesurface for coactivator interaction. (A) AR (PDB, 1T7R), in comparison to (B) ER (PDB, 2J7X),provides more space for interaction with bulky, aromatic residues reported to be selective ARbinders. Charged amino acids considered to make important electrostatic interactions areobserved in parentheses.

Ligand binding to the LBP of the NR LBD, and the subsequent conformational change,including a repositioning of the mobile helix 12 (H12) region, results in coactivator binding tothe NR AF 2. In the case of AR, this binding platform consists of nine amino acids, which areinvolved in forming the hydrophobic interaction with the coactivator protein (Figure 5 2a):V716 and K720 on helix 3 (h3), Q733, M734, I737, and Q738 on helix 4 (h4), and M894, E897,and I898 on H12. Although six of these nine residues are different compared with ER , thegeneral hydrophobic environment does not change, since the amino acid residues remainpredominantly hydrophobic. One exception is E287(ER ) Q738(AR), which results in a netloss of a negative charge. Furthermore, residues with reduced steric bulk, whichaccommodates the greater steric bulk of the aromatic phenylalanines as a unique feature ofthe FxxLF binding motif AR. In particular, residues I898 (in place of M449) and V716 (inplace of I265) provide more room for interaction with the first phenylalanine (Figure 5 7,Phe ). Furthermore, the distance between the charge clamp residues (AR: E897/K720 andER : E448/K269) is 5% greater for AR. This AF 2 shape difference in combination withknowledge acquired from previous ER screening facilitated the design of a set of novel CBIs(Figure 5 1, In Silico Design Rationale).

5.3. Molecular Dynamics Simulations – identification of scaffold

The helical propensity of the designed, proline derived peptides was verified usingmolecular dynamics (MD) simulations (Figure 5 1, I. MD simulation). Implicit solvent wasused starting with an extended initial conformation. In fully unrestrained conditions, alltrajectories were generated and each peptide was simulated for a total of 20 ns at 51.9 °C (325K), in accordance with a similar approach as previously reported to predict the conformationof a miniprotein.32 Analysis was performed on the 20 ns time scale using 20,000 trajectorysnapshots spaced at 1 ps intervals. Secondary structure analysis of each residue as a functionof time was subsequently performed.33,34

A B

Chapter Five

106

Figure 5 3 | Initial MD simulationsMolecular Dynamics (MD) simulations were performed to find the optimal scaffold with respectto helicity. Helicity vs. peptide sequence derived from a 20 ns molecular dynamics simulation:comparison of peptides 1, 2, 3 and 4. The degree of helical content per residue was obtained usingthe ptraj module of AMBER. Finally, 4 consists of an N’ terminal part of peptide 2 and 3 (dashedline) and both the FXXLF motif and C’ terminal part of 1 (black line), with an V P mutation(grey line).

Initially, MD simulations were performed to find the ideal scaffold for introducing both theAR specific FxxLF motif and the helix modulating/controlling prolines. A logical startingpoint was the sequence of the N’ terminal domain of AR (1), an FXXLF bearing sequencecapable of interacting with the AF 2. However, this peptide already showed limited helicalcharacter around Phe , prior to introducing the proline residues (Figure 5 3A). Analternative approach used the natural coactivator SRC1 Box2 as a scaffold with which tointroduce both the central FxxLF motif and the flanking prolines (2 and 3, respectively). Butagain the propensity to form helical structures in this case was not optimal according to theMD calculations. Finally, a combination of the N’ terminal motif of SRC1 Box2 (LTERHPI)bearing both the core motif and C terminus from 1 and additional prolines (FQNLFQSPRE)suggested the best formation of a well defined helix (Figure 5 3, 4).


107

5.4. First generation peptides – the second proline position

First, the optimal position of the second proline was investigated, because structuralanalysis had revealed that, compared to ER, the AF 2 of AR is not only deeper but alsofeatures a larger distance between the two charge clamp residues (Figure 5 1 and 5 2).

Figure 5 4 | Influence of the second proline position on AR bindingMoving the second proline one position in the N’ terminal direction (P+3 P+4) by introducing avaline (V), naturally occurring in AR NTD at that position, leads to enhanced helicity (MD) andAR binding (polarization assay). (A) Polarization competition assay of 5 and 7 and V inserted 6and 8 competing with FLETT 1 (CSSRFESLFAGEKESR, phage display hit AR LBD15). (B) Helicityvs. peptide sequence derived from a 20 ns molecular dynamics simulation: comparison ofpeptides 5 and 6. The degree of helical content per residue was obtained using the ptrajmodule ofAMBER.

Repositioning of the proline ( P VP ) in two different peptides (5 6, 7 8) resulted inboth greater propensity for helix formation and improved binding to AR LBD (Figure 5 4).While 6 and 8 were inactive, the chain extended variants (by one amino acid) were capable ofcompeting with the strong FXXLF reference peptide (5, Ki = 77 M; 7, Ki = 85 M). Togetherwith the FXXLF core motif, this proline at the +4 position – realized by introducing anadditional valine at the +3 position was kept constant in the second set of peptides.

5.5. Influence of first proline position and charge on structure

In this part of the study both the position of the first proline and the influence of the aminoacid after the proline were simultaneously investigated. In a first set of 12 mer peptides,consisting of the sequence X 3X 2X 1FQNLFQSVP (XY = variable positions Y), the proline wasrelocated between position 1, 2 and 3 (PQL~, QPL~,QLP~). A second set of peptides wasprepared keeping the histidine and proline residue at positions 3 and 2, respectively, whilevarying the 1 position (HP X 1~). The choice of residues was as follows: glutamic acid (E)representing negatively charged amino acids; lysine (K) as a positive amino acid, glutamine(Q) with its polar but uncharged side chain properties; and leucine (L) representing

A B

Chapter Five

108

hydrophobic amino acids. All peptides were synthesized on solid phase as 12 mers to enablestructure activity relationship studies.

Table 5 1 | Circular dichroism measurements of AR peptides.

CD spectra of potential AR binders peptide sequence label 222/ 208c

first proline positionAcQLPFQNLFQSVP 9a 0.73AcQPLFQNLFQSVP 10a 0.81AcPQLFQNLFQSVP 11a 0.89

influence of solvent (TFE)AcHPQFQNLFQSVP 12b 0.28bAcHPQFQNLFQSVP 12a 0.82

influence of residue after first proline ( 1 position)AcHPKFQNLFQSVP 13a 0.83AcHPLFQNLFQSVP 14a 1.14AcHPEFQNLFQSVP 15a 0.75

CD spectroscopy measurements were performed under a constant nitrogen flow at 20 °C at 50 M peptideconcentration in phosphate buffer (b) or in additional 30% (v/v) TFE (a, all except 12b). Information about helical

propensity in a semi quantitative manner could be shown by determining the ratio of the characteristicwavelengths at 208 nm and 222 nm: 222nm/ 208nm (c).

The investigation of the secondary structure of the synthesized peptides was determinedby CD spectroscopy. These structural data are consistent with the MD simulation data(Figure 5 5), which show, for instance, the propensity to form longer helices upon gradualrepositioning of the proline from position 1 to position 3 (Table 5 1 and Figure 5 5A, 9 11 ).Furthermore, the observed influence of the helix promoting solvent 1,1,1 trifluoroethanol(TFE), which simulates the proximity of the protein surface, fits coherently with thetheoretical and practical structural analysis. With no observable helical character in wateraccording to MD simulations, or a low 222/ 208 ratio based on CD measurements inphosphate buffer (an indicator for helicity), TFE was observed to cause an increase in thedegree of secondary structure formation (Table 5 1, 12). The result for 14, having the mosthelical character, is coherent for both studies. (Table 5 1, and Figure 5 5B, 13 15).


109

Figure 5 5 | Structural analysis of proline based peptide binders for ARHelicity vs. peptide sequence derived from a 20 ns molecular dynamics simulation. Investigationsconcerning proline position (A, 9 11) and content of 1 position (B, 12 15), respectively. Thedegree of helical content per residue was obtained using the ptrajmodule of AMBER.

5.6. Influence of the first proline position and charge on binding

After structural analysis, the peptides were tested for their ability to bind AR LBD. In acompetitive experimental set up, potential CBIs were made to compete with the fluoresceinlabeled FLETT 1 peptide identified from phage display (Kd = 0.67 0.07 M)15 for bindingto the AR AF 2. Overall, the peptides produced classical sigmoidal curves and behaved ascompetitive inhibitors with Ki values in the range of 1 M to 110 M (compared to theFXXLF featuring peptide with a sequence based on AR NTD, 1, Ki = 0.6 M). Theinvestigations with respect to the proline position N’ terminal of the fixed FXXLF motifidentified position 2 (10) as the optimal position (Ki = 1.5 M). The peptide analogue withproline at 3 position (11) was equally active (Ki = 1.8 M). However location of the proline atthe 1 position, directly prior the FXXLF motif (9) resulted in a significant decrease in bindingaffinity (Ki = 64 M). Keeping the Pro 2 fixed and the simultaneous exploration of variousamino acids residues at the position between Pro 2 and the FxxLF motif, identified bothpositively charged (13) and hydrophobic (14) residues to be beneficial for peptide binding (Ki

= 3.5 M and 4.1 M, respectively). In contrast, polar (12) and negatively charged aminoacids (15) resulted in poor binding affinities (Ki = 42 M and 110 M, respectively). Furtherinvestigations into the most active peptide (10) against other NRs, including estrogenreceptor (ER ) and the retinoic X receptor (RXR ), confirmed the selectivity for AR.Cellular binding with full length AR showed enhanced binding for 10 compared to 1 andARA54, a known AR specific coactivator.

A B

Chapter Five

110

Figure 5 6 | Evaluation of proline based peptidesDepolarization competition assays with fluorescein labeled cofactor (AR: FLETT 1, ER :SRC1 Box2, RXR : D22). (A) Competition assay of peptides with varying proline position (9 11).(B) Competition assay of peptides with different residues at position 1 (12 15).(C) Binding properties of 10 against ER and RXR . D22, LPYEGSLLLKLLRAPVEEV,IC50 = 184 nM; S1B2, SRC1 Box2, SLTERHKILHRLLQEGSPSD, IC50 = 289 nM. (D) Mammalianone hybrid studies on peptides 10 and 13 comparing normalized luciferase activity with knownAR binders, including AR NTD (1) and ARA54. L and S are assigned to the linker between thesequence and bait domains. 13, 10, and 10* have corresponding linker to 1S. 10* is peptide 10with removal of Val+3 resulting in Pro+3. (E) IC50 resulting from a fit of the sigmoid curves. Ki

values calculated using equation 3a,b (Experimental). FXXLF motif highlighted in light grey,flanking prolines underlined and bold; a ± sign is the standard error of the mean (s.e.m.); b nobinding.

peptide sequence name IC50 (±)a[ M]

Ki (±)a[ M]

AcHPNFQNLFQSVP 5 85.10 ± 44.98 38.63 ± 14.14AcHPNFQNLFQSP 6 n.b.b n.b.bAcHKPFQNLFQSVP 7 77.75 ± 35.73 35.58 ± 16.42AcHKPFQNLFQSP 8 n.b.b n.b.bAcQLPFQNLFQSVP 9 140.4 ± 8.374 64.38 ± 3.847AcQPLFQNLFQSVP 10 2.806 ± 0.278 1.148 ± 0.127AcPQLFQNLFQSVP 11 4.244 ± 1.180 1.809 ± 0.542AcHPQFQNLFQSVP 12 93.06 ± 21.01 42.62 ± 10.07AcHPKFQNLFQSVP 13 7.851 ± 0.381 3.466 ± 0.175AcHPLFQNLFQSVP 14 9.237 ± 0.252 4.102 ± 0.116AcHPEFQNLFQSVP 15 241.3 ± 65.13 110.7 ± 29.92

E

A B

C D


111

5.7. Discussion

In this study the AR coactivator interaction was investigated as a potential target to blockAR function. Targeting this hydrophobic interaction instead of the ligand binding pocket(LBP) is a potentially advantageous approach, given that classical antagonists tend to sufferfrom undesired toxicity, in addition to long term treatment resistance caused by pointmutations in the AR LBP, enhanced ligand production, or up regulation and modification ofeither AR itself or AR coactivators and accessory proteins.7,12 For the design of novel ARpeptide inhibitors, we applied knowledge about prolines as helix breakers, obtained fromprevious investigations into ribosome display screening of the ER surface, and combined thiswith the predicted helical propensity of the potential CBIs (a known requirement forefficient binding), as determined by molecular dynamics (MD) simulations. Our resultsdemonstrate the possibility of knowledge transfer within the steroid receptor subfamily ofthe NR superfamily resulting in a proline derived 12 mer peptide endowed with strong andselective binding affinity for AR. The initial evaluation of in silico designed peptides inrespect of their secondary structure eliminates streamlines the development process and inparticular limits the possibility for negative hits at the early stages of the study.The initial position of the prolines in the AR targeting peptides was based directly on the

ER screening investigations. This study exposed flanking prolines at both the 2 and +3position with respect to the central LXXLL motif (chapter 4). Structural analysis by cocrystallization and CD studies connected these positions to N1 and C’, respectively.35 Thefirst choice of scaffold structure for potential AR CBIs design was the N’ terminal AR domain(1), which interacts with the AR AF 2 via an FXXLF motif. However, MD simulationspredicted low helical character in this case, especially in the region of the firstphenylalanine Phe (Figure 5 3A, 7 Phe). The reason for the poor helix propensity could bethe presence of a glycine residue N’ terminal of the FXXLF motif. Previous studiesinvestigating the helical propensity of different peptides and proteins have shown thatglycines aside only from proline – possess the lowest helix propensity.36 The formation ofhelices is an interplay between the loss of side chain configurational entropy and the gain

of enthalpy through mainly hydrogen bond formation between i and i + 4 amino acids.37,38

Apart from the side chain based entropy difference between the helical and coiled state(helix coil transition) there are also different enthalpic contributions from the residues.38

Thus, in the case of glycine, aside from the high entropic costs through fixing of the backbonedihedral angles in an helix, the enthalpic contribution turns out to be the lowest of allresidues, which is again a combination of the poor solvent accessibility of the backbone andthe lack of possible stabilizing van der Waals interaction between amino acid side chains.39,40

The next step in designing proline derived AR inhibitor peptides focused on theintroduction of both the specific FXXLF motif and the flanking prolines into the SRC1 Box2coactivator sequence also part of the aforementioned ER ribosome display study which

Chapter Five

112

resulted in a PXFXXLFXXP motif (Figure 5 3C, 3). The introduction of the aromatic aminoacid residues caused a destabilization of the helical secondary structure, compared to itsequivalent LXXLL bearing sequence (Figure 5 3B, 2). Compared to 1, however, theN’ terminal region between Phe and Pro 2 showed improved helicity, while the totalhelicity, especially at the C terminus (Phe /Pro+3) was observed to decrease. Indeed,phenylalanines display a lower propensity to form helices compared to leucines.36

Nevertheless, this dramatic loss of overall helical character proved to be an unexpectedoutcome of these investigations.A logical next step in the design, therefore, was to combine the most suitable characteristics

of both sequences, since one (3) possessed a well defined structure in the region surroundingPhe , while the other (1) around Phe . The 11 mer resulting from this new design consistedof the following sequence: H P I F Q N L F Q S P. MD simulation predicted in this case agood helical propensity with prolines in N1 (beginning of helix) and C’ position (twopositions after the helix) (Figure 5 3).With this scaffold sequence in hand, the ideal position of the second proline was

investigated with respect to helicity and finally AR binding. This first generation AR CBIswere inactive with the second proline located at position +3 (Pro+3), however, introduction ofa valine at position 3 (Val+3) and consequent repositioning of the proline (Pro+4) resulted inenhanced binding properties. This result would suggest that AR prefers longer helices forcoactivator binding compared to ER. Indeed, analysis of the AF 2 of AR in comparison withER reveals the slightly greater distance between the two charge clamp residues glutamicacid and lysine (Figure 5 7). This difference of 0.8 Å could provide sufficient space for anadditional amino acid in the helical segment spacing in between the two charged amino acids(3.6 Residues are needed for one helix full turn of 5.4 Å, resulting in a 1.5 Å contributionper amino acid).41 The improved binding properties are also a function of helix stability.MD data showed a significant increase in helical character upon a repositioning of theproline from +3 to +4 (Figure 5 4). Thus, the introduction of the valine in position +3 (Val+3)plays a crucial role in stabilizing the helix.


113

Figure 5 7 | Comparison of AF 2 (AR, ER ) and influence of valine on the helix stabilization(A) Comparison of distances between the side chains of charge clamp residues for ER(E542/L362, PDB, 2J7X) and AR (E897/L720; PDB, 1T7R), which are responsible among otherresidues – for AF 2 interacting peptides. (B) Superimposition of the predicted structure of 10 (MDsimulation) with AR LBD FXXLF coactivator complex (PDB, 1T7R). 10: Phe / , first and lastphenylalanine in FXXLF; L , leucine in FXXLF. Pro 2/+4, Val+3, flanking residues.

Hydrophobic interactions are important in protein folding and for protein proteininteractions (PPIs). The driving force for the hydrophobic effect is the reduction ofhydrophobic residues exposed to polar water.42 Since helices are mainly stabilized byelectrostatic interactions, such as hydrogen bond formation between i and i + 4 residues,hydrophobic interactions of side chains receive less attention, and the degree with whichthey contribute to helix stability is the subject of ongoing discussions. However, studies onalanine based peptides have shown that hydrophobic residues at the i and i + 4 positionsexert a positive effect on the helical nature of the peptide.43,44 Even though the two sidechains at i and i + 3 sit close to one another, the side chain orientations of the i, i + 4 residuesare more suitable for hydrophobic interactions, including van der Waals and / stacking.37,45

Thus the significantly increased binding properties of peptides 5 and 7 over 6 and 8 are afunction of enhanced helix stability through additional interactions between the hydrophobicresidues Leu Val+3 (Figure 5 7B, C H H C , 3.4 Å). This contribution to helix stabilityseems to be necessary due to the lower helix propensity of phenylalanines (Phe /Phe ofFXXLF) compared to leucines (Leu /Leu of LXXLL), which are used in ER binding.Furthermore, the introduction of additional hydrophobic residues (Val+3) is again onlypossible due to the greater distance between the AR charge clamp residues E897/L720(compared to E542/L362 in ER ), thus providing space for a more extended helix (Figure5 7A).In the second library of AR CBIs the core structure (FXXLF) and the C’ terminal Val Pro

motif remained unchanged, while the N’ terminal residues surrounding the first proline(positions 1, 2, and 3) were varied. First, the ideal position of the first proline wasinvestigated. CBIs were designed with varying proline positions and two other neutral

A B

Chapter Five

114

amino acids of high helical propensity – namely the polar glutamine and hydrophobicleucine – which most likely exert little effect on the binding affinity of the peptide inhibitor.The circular dichroism (CD) data support the predictions made by MD simulations, in thatthe movement of the proline from position 1 through 2 to 3 resulted in a more extendedhelix ( mre,222nm/ mre,208nm: 11 > 10 > 9, Table 5 1). Competitive binding studies showed that Pro2 has similar binding compared to Pro 3, both with strong inhibitory properties (10, Ki = 1.1;11, Ki = 1.8 M). Co crystal structures with other coactivators peptide sequences have shownthat the interaction with the backbone and the charge clamp does not occur at the terminus ofthe helix, but instead at the amide bond located between the first two amino acids (here, ProGln, if Pro 2 = N1) in the case of the interaction with the charge clamp residue E897. At theopposite end of the charge clamp, K720 interacts with the amide bond situated between thefourth and third last amino acid residues (here, Leu Phe, if Pro+4 = C’). Combining thisknowledge with a helix contribution of 1.5 Å per residue and taking a typical hydrogen bonddistance of 2.7 3.1 Å into account, results in a theoretical, optimal helix length of 9.8amino acids. Peptides with proline at position 2 or 3 are matching this prediction with ahelix length of 9 and 10 residues, respectively (assuming prolines in N1 and C’ position)(Figure 4 8, Figure 5 7B). Thus, the results from the binding studies confirm the theoreticalideal length defined by the flanking prolines, and provide a point of entry to address NRsubtype selectivity based on helix length recognition. Interestingly, cluster analysis of theAR specific motifs (Figure 5 8B) showed a preference for proline at the 2 position.After identifying the optimal position of the first proline, the position between this residue

and the FXXLF motif was next investigated (position 1): H P (X 1) F Q N L F Q S V P.Lysine and glutamic acid were chosen, representing positively (13) and negatively charged(15) residues, respectively. Furthermore, a hydrophobic (14) and polar (12) amino acid wereintroduced. Perhaps unsurprisingly, 15, bearing a glutamic acid at the 1 position, displayedthe weakest binding properties. Crystal structure analysis of AR LBD revealed that thecharge clamp residues, E897 and K720, conserved across all NRs form hydrogen bondswith the backbone of coactivators, and in this way determine the size of the axial length.46

Together with the adjacent residues E893 and E709, E897 forms a cluster of negative chargeon one side of the AF 2. Introduction of a negative amino acid at the positions adjacent to thefirst proline (15) resulted in electrostatic repulsion and thus in a weaker binding affinity. Onthe other hand, mutation to a positively charged lysine (13) resulted in significantlyenhanced binding (Ki = 3.5 M). These results are in line with the LBD screening of both ARand also ER, which predominantly identified positively charged amino acids amino terminalto FXXLF or LXXLL, respectively.14,15 Furthermore, the cluster analysis of nuclear receptorcoactivators, including the p160 family, and specifically androgen receptor associated (ARA)protein family, including ARA54/55/70, show the occurrence of positive residues aminoterminal to the LXXLL and FXXLF motif (Figure 5 8). Although, negative amino acids


115

residues are also found in this region, especially in the case of the ARA family, they arealways in direct proximity to a neutralizing positive charged amino acids, while positiveresidues typically stand alone.18

The effects of introducing polar residues (12 and 5) compared to the hydrophobic leucine(14) were less dramatic. Yet the hydrophobic residue still invoked an order of magnitudegreater inhibition (Ki = 4.1 vs. 38.6/42.6 M). This is perhaps as expected, since this specificPPI is based on hydrophobic interactions, including leucines or phenylalanines. In general,hydrophobic residues are important for the interior of both protein monomers and proteinprotein interfaces, having a large positive entropic effect, which is directly connected to therelease of water molecules from the cavities of hydrophobic surfaces upon protein foldingand PPI, respectively. Although the influence on PPIs is not as strong as for protein folding,it remains the principle driving force. However, salt bridge formation, including hydrogenbonding, is also contributing to these interactions because of the unfavorable, hydrophobicsurfaces of unassociated monomers exposed to the polar environment and the greaterpossibilities for selectivity necessary in the case of PPIs.11,47

Figure 5 8 | Cluster analysis of NR and AR coactivators(A) Cluster analysis of NR coactivators featuring the LXXLL motif (n = 40) and (B) coactivatorinteracting nearly exclusively with AR via the FXXLF motif (n=5).Weblogo by Crooks et al.48

In addition, mutation of the histidine at position 3 of the peptide (14, His 3), which playeda major role in ER binding (chapter 4), to an uncharged glutamine (10, Gln 3) led toenhanced inhibition of AR coactivator binding (Ki = 4.1 to 1.1 M). Comparison of the AF 2 ofboth NRs reveals that ER has a negatively charged residue E287 to which the positivelycharged His 3 is pointing towards, while AR possesses a neutral amino acid (Q738) at theequivalent position. Thus the difference in residue preference between these two NRs couldbe explained by the subtraction of charge in this part of the AF 2. Histidine (positivelycharged and a hydrogen bond donor) and glutamic acid (negatively charged and a hydrogenbond acceptor) are good interaction partners in terms of salt bridge formation, through acombination of hydrogen bonding and electrostatic interactions. On the other hand,glutamines are both hydrogen donor and acceptor, and thus well suited for hydrogen bondformation with each other.49

Subsequently, the most potent inhibitor from this study was tested for its selectivity. Thus,both another member of the subclass of steroid receptors (NR3) AR and a member of the

A B

Chapter Five

116

class of retinoid receptors (NR2) – RXR were tested for their binding properties to bind 10.In both cases no binding could be detected, suggesting that the peptides found during thisline of investigation show exclusive inhibition for AR. Final cellular studies confirmed thegood binding properties of 10 and 13. Interestingly, although both peptides showed weakerbinding compared to 1 in the biochemical assay (polarization assay), in this mammalian onehybrid set up 13 showed a similar and 10 even four fold enhanced ability to activateluciferase expression (Figure 5 6). This can be explained by the fact that in this binding assayreceptor activity can be obtained by the intact protein compared to only the LBD in thebiochemical assays. And especially for AR it is known that the N/C interaction between NTDand LBD plays a major role in the overall transcriptional activity.18,50,51

5.8. Conclusion

In conclusion, a successful conceptual approach for developing selective AR CBIs has beendescribed. The rational in silico design was based on knowledge from investigations into thePPI between the AR related ER and its coactivators in which prolines were identified as helixregulators (breaking and stabilization) resulting in enhanced inhibitory properties. MDsimulation studies revealed that phenylalanines present in the AR exclusive FxxFL bindingmotif contribute less to the formation of a well structured helix compared to leucines(LXXLL motif). Thus, for the introduction of both FXXLF motif and flanking prolines thescaffold sequence had still to be optimized. A combination of the NCoA1 Box2 sequence,known to bind various NRs, and a sequence based on the N’ terminal domain of AR (ARNTD) showed the best propensity to form an helix and was therefore used as the basicsequence. This combination of the unique AR binding ability of phenylalanine rich motifswith helix breaking prolines resulted in short peptides with moderate binding properties(Kd/ki valuesin the high M range) for the AR LBD. Further optimization focusing on thesecond proline position (Pro+4 >> Pro+3), the first proline position (Pro 2 > Pro 3 >> Pro 1),the 1 position between Pro 2 and FXXLF motif (K > L >> Q > E), and position 3 prior to Pro2 (Q > H) culminated most notably in a 12 mer peptide with promising CBI properties (10,Ki = 1.1 M). Cellular studies involving intact AR identified the ability of 10 to overpowerknown AR AF 2 interaction partners, such as the AR NTD or AR associated protein 54(ARA54). Thus, this peptide based inhibitor can be seen as a lead compound targeting theAR AF 2 surface and thus AR activity in preferential manner, based exclusively on naturalamino acids and without recourse to additional chemical modifications. The fact that itinterferes one step downstream of classical ligand competing drugs opens new possibilitiesin drug selectivity and treatment of both androgen resistant prostate cancer and diseasesbased on ligand independent misregulation.


117

5.9. Experimental

GeneralMD simulation studies were performed by Dr. Lidia Nieto. The final cellular investigations wereperformed by Dr. Ingrid J. de Vries van Leeuwen.

If not stated otherwise the following contents, descriptions, protocols, and distributer are valid:chemicals were ordered from SIGMA ALDRICH. Proteins were handled at 4 °C. Standard culturingmedium is lysogeny broth (LB) medium (10 g peptone, 10 g NaCl, 5 g yeast, 1 l water, autoclaved).Peptide sequences:

SRC1 Box2, SLTERHKILHRLLQEGSPSD, LXXLL motif, natural NR coactivator (Ncoa1692 711)FLETT 1, CSSRFESLFAGEKESR, FXXLF motif, phage display hit AR LBD15

D22, LPYEGSLLLKLLRAPVEEV, LXXLL motifphage display hit ER LBD14

AR NTD, KTYRGAFQNLFQSVRE, motif FXXLF, NTD of AR, binding to AR AF 2 (AR17 32)ARA54, NDPGSPCFNRLFYAVDVDD, motif FXXLF, specific AR coactivator (RNF14447 465)

Molecular Dynamics SimulationAll MD simulations were carried out using the Assisted Model Building with Energy Refinement(AMBER) suite of programs, and the ff03 force field.52 An implicit solvent was used via the GeneralBorn solvation method (IGB 5), as implemented in AMBER.53 Starting with an extended initialconformation, built by the LEaP module of AMBER for each peptide, all MD simulations were fullyunrestrained and all trajectories were generated using the sander program in the AMBER 9 package.Each peptide was simulated for a total of 20 ns at 51.9 °C, following a similar approach as previouslyreported to predict the conformation of a miniprotein.32 Analysis was performed on the 20 ns using20,000 trajectory snapshots spaced every 1 ps. A snapshot with the lowest potential energy across thesimulation was chosen as representative structure for each peptide. The secondary structure of eachresidue as a function of time was subsequently analyzed utilizing the STRIDE secondary structureassignment algorithms as implemented in VMD.33,34 All MD simulations were performed by Dr. LidiaNieto.



construct vector tag protein originAR LBD pGEX KG GST hAR LBD664 919 BAYER HEALTHCARE PHARMACEUTICALS



Plasmids with the desired protein construct (Table 5 2) were transformed into E. coli BL21 cells.Emerging colonies were cultured over night in 25 ml LB medium with 100 g/ml ampicillin at 37 °C.This preculture was added to 2 l phosphate buffered, content rich terrific broth (TB) medium (12 gpeptone, 24 g yeast, 4 ml glycerol, 0.17 M KH2PO4, 0.72 M K2HPO4, 1 l water, autoclaved) withampicillin and incubated at 37 °C until an optical density at 600 nm (OD600) of 0.6 1.0. Addition of100 M Isopropyl D 1 thiogalactopyranoside (IPTG) induced expression of the protein of interest.For stabilization 10 M of ligand (ER, estradiol | AR, dihydrotestosteron | RXR, agonist LG) was

Chapter Five

118

added. After incubation for 20 h at 15 °C the cells were harvested by centrifugation (15 min at 7,000rpm) and the resulting pellet was stored at 80 °C until further usage.If not stated otherwise, the protein purification was performed at 4 °C. The pellet was resuspended inlysis buffer (LyB 1: 1x PBS tablet, CALBIOCHEM, 370 mM NaCl, 10% (v/v) glycerol, 0.1 mM TCEP,1 mM PMSF, 1 g/ml DNAseI, 10 M ligand, pH 8.0, His6 tag: additional 40 mM imidazole) and lyzedwith the Emulsi Flex C3 homogenizer (2 passes of 150,000 kPa, AVESTIN INC.). After anothercentrifugation step (40 min, 20,000 rpm) the supernatant was provided on the respective affinitychromatography column (GST tag: Protino GST/4B, 1 ml, MACHEREY NAGEL | His6 tag: His BindResin, 3 ml, NOVAGEN), washed with wash buffer (WaB 1: GST tag: 1x PBS tablet, CALBIOCHEM, pH 7.3| His6 tag: 1x PBS tablet, CALBIOCHEM, 370 mM NaCl, 40 mM imidazole, 10% (v/v) glycerol,0.1 mM TCEP, 10 M ligand, pH 8.0) and eluted with competing molecules (GST tag: 50 mM Tris,10 mM glutathione | His6 tag: WaB 1 plus 100 mM / 250 mM / 500 mM imidazole). Fractions wereanalyzed with SDS PAGE and optional with LC MS.

Solid Phase Peptide Synthesis & PurificationThe synthesis of all peptides was performed on solid support, using the Fmoc strategy. If not statedotherwise, 200 mol (1 eq.) of Rink Amide MBHA resin (0.59 mmol/g, Novabiochem) was swollen for30 min in N methylpyrrolidone (NMP). The protective Fmoc group was removed by incubation with20% piperidine in NMP (2x 5min). Subsequently the amino acids (AA, 4 eq.) were activated using2 (1H Benzotriazole 1 yl) 1,1,3,3 Tetramethyluronium hexafluorophosphate (HBTU, 4 eq.), and N,NDiisopropylethylamine (DIPEA 8 eq.). Single coupling was performed by shaking 45 min at roomtemperature. After finishing the desired peptide sequence, the resin was washed 3 times alternatelywith diethylether (DEE) and dichlormethane (DCM) and dried under vacuum for 10 min. The sidechain protection group and the cleavage of the peptide sequence from the resin were simultaneouslyachieved by the usage of mixture of trifluoroacetic acid, triisopropylsilane and water (TFA/TIS/H2O,95/2.5/2.5) for 3 h. The free peptide was precipitated by adding it drop wise to ice cold DEE. A finalcentrifugation step (2.000 rpm, 10 min) with subsequent washing (ice cold DEE), uptake in water andlyophilization results in the crude peptide. Purification was accomplished by reverse phase HPLC ona Alltima™ HP C18 column (125x20 mm, Alltech). Water (0.1% TFA) was used as a polar phase, addingdifferent amounts of an apolar acetonitril (ACN, 0.1% TFA) phase. A linear gradient (20 ml/min) wasoptimized for each peptide to a 5% range, for instance 35 40% ACN. The purity of the peptide wasdeterminate at an analytical LC MS using GraceSmart RP18 (50x2.1mm, Grace, 3u, 120A) columnFinally, the peptides were lyophilized and stored at 80 °C.

Fluorescence Polarization and Depolarization assay55,56

Both fluorescence polarization (FP) and fluorescence depolarization (FDP) experiments weremeasured in black 384 well plates (PERKIN ELMER 384 F) on a Safire2 (TECAN) plate reader. The finalvolume of one well was 30 l containing constant concentrations of ligand (5 M) and Fluoresceinlabeled peptide (0.1 M, ER: NCoA 1/SRC1683 701, LTERHKILHRLLQEGSPSD; AR: FLETT 1,CSSRFESLFAGEKESR, phage display hit AR LBD;15 RXR : D22, LPYEGSLLLKLLRAPVEEV,INVITROGEN, phage display hit ER LBD, good binding properties RXR 14). In case of FP assay theprotein was sequentially diluted (24 steps) in TR FRET coregulator buffer E (INVITROGEN). On theother hand, during FDP assays the protein concentration was kept constant (ER: 0.4 M; RXR, AR:1 M), varying the concentration of the potential peptide based binder. Briefly, in 96 well platesdilution series were prepared in 55 l and filled up with to 110 l with a 2X master mix solutioncontaining all non varying components. Finally three times 30 l was transferred to 384 well plates,


119

centrifuged (1,000 rpm, 2 min) and incubated for 1 h at 4 °C. Plates were measured 50 times at 30 °C(excitation 470 nm, emission 519 nm). Polarization (in [mP]) was plotted against the concentrationof either the protein (FP) or the test peptide (FDP), whereby each data point represents an average of 3experiments.The dissociation constant Kd of the protein peptide complex (FP) was calculated using equation 1:

where P is the measured polarization value (in mP), P0 is the polarization of the free fluorescentligand, Pfin is the polarization of the bound ligand, A0 is the total concentration of fluorescent peptide,and B0 is the protein concentration.In the competitive setup the half maximal inhibitory concentration ( ) was calculated with themeans of equation 2a and 2b:

where is the bottom asymptote, is the top asymptote, and the hill slope (steepness of thecurve).The inhibitory constant Ki of the different peptides could be determinated with the value of theprotein and equation 3a and 3b.

where Kd is the dissociation constant of the fluorescein labeled peptide protein complex (equation 1),y0 is the initial bound to free concentration ratio for labeled peptide, [AB] is the concentration ofprotein peptide complex, [A] the concentration of unbound peptide, and [B] the concentration ofprotein (ER).

Circular Dichroism spectroscopyFar UV Circular Dichroism (CD) spectroscopy measurements were performed under a constantnitrogen flow at 20 °C using 50 M of the peptide in the presence of 30% (v/v) trifluoroethanol (TFE),if not stated otherwise. As a baseline a peptide free buffer (30% TFE) was measured under identicalconditions. The spectrum was recorded from 250 nm to 185 nm using a JASCO 815 spectrometer.Quartz cuvettes with path lengths of 1 mm or 0.2 mm (HELLMA ANALYTICS) were employed with adata pitch of 0.5 in a continuous mode, a scan speed of 20 nm/min with a response time of 2 s and abandwidth of 0.5 nm. The graphs are representing an average of five scans. The observed ellipticity(degrees, in [mdeg]) was converted to mean residue ellipticity (in [deg cm2/dmol]) usingequation 3.

Chapter Five

120

Where is the molecular mass of the peptide (in [Da]), is the number of amino acids in the peptide,is the path length (in [cm]), and is the peptide concentration (in [g/ml]). Information about helical

propensity in semi quantitative manner could be shown by the ratio of the characteristic wavelengthsat 208 nm and 222 nm: .

Mammalian One Hybrid assayAll Mammalian One Hybrid assays (M2H) were performed in human osteosarcoma cells (U2 OS57)seeding 30,000 cells/well in a 24 well plate (BD Biosciences). Plasmids are either from MammalianTwo Hybrid Assay Kit (AGILENT) or cloned by Dr. Ingrid J. de Vries van Leeuwen. Culturingconditions were 37 °C and 5% CO2. Cells were transfected with following amounts of DNA/well usinglinear polyethylenimines (PEI, POLYSCIENCES): (1) 40 ng of the bait construct pCMV BD PEPcontaining peptide of interest (PEP) fused to the DNA binding domain of the GAL4, (2) 40 ng of thetarget construct pDNA3.1 AR, (3) 200 ng of the reporter plasmid pFR Luc with the synthetic promoterwith five tandem repeats of the GAL4 binding sites that control expression of the Photinus pyralisfirefly luciferase, and (4) 3.2 ng of control plasmid Renilla Luc, expressing independently Renillareniformis luciferase. After 16 h incubation cells were treated with 10 nM estradiol and after additional24 h incubation the interaction was determined with a Dual Luciferase Reporter Assay (PROMEGA),according to manufacturer’s instruction. The luminescent intensities were recorded on a Synergy HTMulti Mode Microplate Reader (BIOTEK). The interaction dependent Firefly luciferase signal was firstnormalized over the independent Renilla luciferase signal and subsequently normalized over the ratioof untreated cells.


121

5.10. References

1. M. Bhasin and G. P. S. Raghava, J Biol Chem, 2004, 279, 23262–23266.2. E. J. Small and N. J. Vogelzang, J Clin Oncol, 1997, 15, 382–388.3. T. R. Brown, Endocrinology, 2004, 145, 5417–5419.4. F. Zhu, Z. Shi, C. Qin, L. Tao, X. Liu, F. Xu, L. Zhang, Y. Song, X. Liu, J. Zhang, B. Han, P. Zhang,

and Y. Chen, Nucleic Acids Res, 2012, 40, D1128–1136.5. C. E. Bohl, W. Gao, D. D. Miller, C. E. Bell, and J. T. Dalton, Proc Natl Acad Sci USA, 2005, 102,

6201–6206.6. M. E. Salvati, A. Balog, D. D. Wei, D. Pickering, R. M. Attar, J. Geng, C. A. Rizzo, J. T. Hunt, M. M.

Gottardis, R. Weinmann, and R. Martinez, Bioorg Med Chem Lett, 2005, 15, 389–393.7. M. E. Taplin, Nat Clin Pract Oncol, 2007, 4, 236–244.8. A. K. Ramani, R. C. Bunescu, R. J. Mooney, and E. M. Marcotte, Genome Biol, 2005, 6, R40.9. J. Voisey and C. P. Morris, Curr Drug Discov Technol, 2008, 5, 230–235.10. B. Hughes, Nat Biotechnol, 2011, 29, 296–296.11. E. Valkov, T. Sharpe, M. Marsh, S. Greive, and M. Hyvönen, Top Curr Chem, 2012, 317, 145–179.12. V. Buzón, L. R. Carbó, S. B. Estruch, R. J. Fletterick, and E. Estébanez Perpiñá,Mol Cell Endocrinol,

2012, 348, 394–402.13. G. Jenster, H. A. van der Korput, J. Trapman, and A. O. Brinkmann, J Biol Chem, 1995, 270, 7341–

7346.14. C. y Chang, J. D. Norris, H. Grøn, L. A. Paige, P. T. Hamilton, D. J. Kenan, D. Fowlkes, and D. P.

McDonnell,Mol Cell Biol, 1999, 19, 8226–8239.15. E. Hur, S. J. Pfaff, E. S. Payne, H. Grøn, B. M. Buehrer, and R. J. Fletterick, PLoS Biol, 2004, 2, e274.16. B. He, J. A. Kemppainen, and E. M. Wilson, J Biol Chem, 2000, 275, 22986–22994.17. H. J. Dubbink, R. Hersmus, A. C. W. Pike, M. Molier, A. O. Brinkmann, G. Jenster, and J.

Trapman,Mol Endocrinol, 2006, 20, 1742–1755.18. B. He, J. T. Minges, L. W. Lee, and E. M. Wilson, J Biol Chem, 2002, 277, 10226–10235.19. L. Wang, C. L. Hsu, J. Ni, P. H. Wang, S. Yeh, P. Keng, and C. Chang,Mol Cell Biol, 2004, 24, 2202–

2213.20. B. He, R. T. Gampe Jr., A. J. Kole, A. T. Hnat, T. B. Stanley, G. An, E. L. Stewart, R. I. Kalman, J. T.

Minges, and E. M. Wilson,Mol Cell, 2004, 16, 425–438.21. C. Chang, J. Abdo, T. Hartney, and D. P. McDonnell,Mol Endocrinol, 2005, 19, 2478–2490.22. H. Göksel, D. Wasserberg, S. Möcklinghoff, B. V. Araujo, and L. Brunsveld, Bioorg Med Chem, 2011,

19, 306–311.23. J. R. Gunther, A. A. Parent, and J. A. Katzenellenbogen, ACS Chem Biol, 2009, 4, 435–440.24. P. Axerio Cilies, N. A. Lack, M. R. S. Nayana, K. H. Chan, A. Yeung, E. Leblanc, E. S. T. Guns, P. S.

Rennie, and A. Cherkasov, J Med Chem, 2011, 54, 6197–6205.25. L. Caboni, G. K. Kinsella, F. Blanco, D. Fayne, W. N. Jagoe, M. Carr, D. C. Williams, M. J. Meegan,

and D. G. Lloyd, J Med Chem, 2012, 55, 1635–1644.26. B. Vaz, S. Möcklinghoff, S. Folkertsma, S. Lusher, J. de Vlieg, and L. Brunsveld, Chem Commun

(Camb), 2009, 5377–5379.27. M. D. Seoane, K. Petkau, B. Vaz, S. Mocklinghoff, S. Folkertsma, L. G. Milroy, and L. Brunsveld,

Med Chem Commun, 2012.28. T. Phan, H. D. Nguyen, H. Göksel, S. Möcklinghoff, and L. Brunsveld, Chem Commun (Camb), 2010,

46, 8207–8209.29. C. Bracken, J. Gulyas, J. W. Taylor, and J. Baum, J Am Chem Soc, 1994, 116, 6431–6432.30. A. Patgiri, K. K. Yadav, P. S. Arora, and D. Bar Sagi, Nat Chem Biol, 2011, 7, 585–587.31. J. Hanes and A. Plückthun, PNAS, 1997, 94, 4937–4942.32. C. Simmerling, B. Strockbine, and A. E. Roitberg, J Am Chem Soc, 2002, 124, 11258–11259.33. D. Frishman and P. Argos, Proteins, 1995, 23, 566–579.

Chapter Five

122

34. W. Humphrey, A. Dalke, and K. Schulten, J Mol Graph, 1996, 14, 33–38, 27–28.35. S. Kumar and M. Bansal, Proteins, 1998, 31, 460–476.36. C. N. Pace and J. M. Scholtz, Biophys J, 1998, 75, 422–427.37. T. P. Creamer and G. D. Rose, Protein Sci, 1995, 4, 1305–1314.38. J. M. Richardson, M. M. Lopez, and G. I. Makhatadze, Proc Natl Acad Sci USA, 2005, 102, 1413–

1418.39. P. Luo and R. L. Baldwin, Proc Natl Acad Sci USA, 1999, 96, 4930–4935.40. D. N. Ermolenko, J. M. Richardson, and G. I. Makhatadze, Protein Sci, 2003, 12, 1169–1176.41. L. Pauling and R. B. Corey, Proc Natl Acad Sci USA, 1951, 37, 729–740.42. L. Lins and R. Brasseur, FASEB J, 1995, 9, 535–540.43. S. Padmanabhan and R. L. Baldwin, J Mol Biol, 1994, 241, 706–713.44. J. S. Albert and A. D. Hamilton, Biochemistry, 1995, 34, 984–990.45. Y. K. Cheng and P. J. Rossky, Nature, 1998, 392, 696–699.46. R. T. Nolte, G. B. Wisely, S. Westin, J. E. Cobb, M. H. Lambert, R. Kurokawa, M. G. Rosenfeld, T.

M. Willson, C. K. Glass, and M. V. Milburn, Nature, 1998, 395, 137–143.47. C. J. Tsai, S. L. Lin, H. J. Wolfson, and R. Nussinov, Protein Sci, 1997, 6, 53–64.48. G. E. Crooks, G. Hon, J. M. Chandonia, and S. E. Brenner, Genome Res, 2004, 14, 1188–1190.49. I. K. McDonald and J. M. Thornton, J Mol Biol, 1994, 238, 777–793.50. E. Langley, Z. X. Zhou, and E. M. Wilson, J Biol Chem, 1995, 270, 29983–29990.51. E. M. Wilson, in Androgen Action in Prostate Cancer, eds. J. Mohler and D. Tindall, Springer US,

2009, 241–267.52. Y. Duan, C. Wu, S. Chowdhury, M. C. Lee, G. Xiong, W. Zhang, R. Yang, P. Cieplak, R. Luo, T.

Lee, J. Caldwell, J. Wang, and P. Kollman, J Comput Chem, 2003, 24, 1999–2012.53. A. Onufriev, D. Bashford, and D. A. Case, Proteins, 2004, 55, 383–394.54. S. Folkertsma, P. I. van Noort, A. de Heer, P. Carati, R. Brandt, A. Visser, G. Vriend, and J. de

Vlieg,Mol Endocrinol, 2007, 21, 30–48.55. P. Wu, M. Brasseur, and U. Schindler, Anal Biochem, 1997, 249, 29–36.56. F. Perrin, J Phys Radium, 1926, 7, 390–401.57. Y. Liang, D. F. Robinson, J. Dennig, G. Suske, and W. E. Fahl, J Biol Chem, 1996, 271, 11792–11797.

CHAPTER SIX 4 Modulation of retinoid X receptor activity by biaryl

natural products

Abstract. The retinoid X receptor (RXR) is an important transcription factor involved in theregulation of gene networks connected to cell growth, cell differentiation, and cell death.Misregulation of RXR or pathways that are based on RXR heterodimer formation with othernuclear receptors (NRs) can lead to various malignancies, such as cancer, cardiovascular andinflammatory diseases. Small molecules targeting the receptor are used in the treatment ofthose diseases. Natural products (NPs) are privileged structures for the discovery of newdrugs since they are intrinsically favored to bind to enzymes or protein receptors.Furthermore, NPs typically have biological relevance and cell permeability, which areadvantageous properties for the development of novel orally active drug therapies. Thus,microarray based nuclear receptor coregulator interaction profiling was performed to screena small compound library of biaryl NPs originating from the Magnolia species for potentialRXR binding. Magnolol and 4’ O methyl honokiol featured (partial) agonism functionality.The microarray profiling assay additionally identified honokiol itself to featurenon competitive antagonism through allosteric RXR binding. Both the lipophilic side chaingroups and the hydroxyl functionality of these NPs were shown to be important aspects oftheir activity. Modifications to the two lipophilic groups of honokiol, towards an improvedmimic of the structure of leucine side chains of coactivator peptides, resulted in an improvedability to inhibit the RXR–coactivator interaction in the profiling assay. This data incombination with the structural similarity with known NR CBIs identifies honokiol as thefirst NP CBI.

Chapter Six

124

6.1. Introduction

Nuclear receptors (NR) are ligand activated gene transcription factors that determine ahost of fundamental (patho) physiological processes including embryonic development,diabetes and cancer.1 Ligand binding to the lipophilic NR ligand binding pocket (LBP)induces a protein conformational change, which leads to the recruitment of co activatorproteins as a prerequisite for the orchestrated assembly of the transcriptional machinery.2

Therefore, NRs represent important drug targets.3 Screening natural products (NPs) is onepossibility to discover new compounds that can function as lead structures in thedevelopment of drugs for the treatment of human diseases.4 NPs offer a rich source ofstructurally diverse, biologically relevant and cell permeable small molecules or scaffoldstructures, which achieve high hit rates across a broad range of in vitro and cell basedscreens compared with scaffolds based on combinatorial chemistry.5 Indeed, the success ofNPs translates at the clinical level, where a high percentage of marketed drugs are eitherwhole natural products, natural product analogues, or natural product like.4 The privilegedstatus of NPs has even been attributed to the notion that, as metabolites of conservativebiosynthetic processes, they are evolutionarily pre disposed to re interact with protein basedtargets.6–11

In the field of NRs several NPs have been discovered to have NR interacting properties,aside from the native NR ligands such as steroids. One of the best characterized NR targetingNP is the isoflavone genistein. It was isolated from soybeans, directly interacts with estrogenreceptors (ER , NR3A1; ER , NR3A2) and reminiscent of the endogenous ligand17 estradiol. It is used as dietary supplement to ease menopausal symptoms.12,13

Guggulsterone isolated from the guggul tree14,15 has antagonistic properties for thefarnesoid X receptor (FXR, NR1H4), is similar to androstenedione, a precursor for estrogensand androgensin steroidogenesis,16 and is marketed in India as an antihyperlipoproteinemicdrug.17,18. Interestingly, Swinhosterol B isolated from the sponge Theonella swinhoei featuresa dual function, videlicet agonism in terms of PXR (NR1I2) and antagonistic properties forFXR (NR1H4).19 Recently, several NPs were discovered that block androgen receptor(AR, NR3C4) transcriptional activity in prostate cancer, namely epigallocatechin 3 gallatefrom green tea,20 atraric from plant Pygeum africanum,21 lupeol found in fruits vegetables andmedicinal plants,22 and niphatenone B from the marine sponge Niphates digitalis.23

Interestingly, the latter binds to the activation function 1 (AF 1) of the N’ terminal domain ofAR instead of the LBP of the ligand binding domain. Advantages of such alternative bindingare both the circumvention of hormone resistance and the reduction in side effects due tocombinatorial treatment involving lower doses of LBP targeting antagonists.24

These benefits have also stimulated investigations of activation function 2 (AF 2) – thebinding platform for coactviators – as a target for small molecule intervention instead of the

Modulation of retinoid X receptor activity by biaryl natural products

125

LBP. A successful strategy for designing coactivator binding inhibitors (CBIs) has been tomimic the structural features of peptides commonly shared across all NR coactivators(peptidomimetics). Pyrimidine scaffolds mimicking the i, i + 3, i + 4 arrangement of LXXLL(ER) or FXXLF (AR) were developed to selectively target ER and AR, respectively (Figure 6 1,A and B).25–28 Other studies have indentified biaryl based small molecules, such aspyridylpyridones (Figure 6 1C)29 or biphenyls (Figure 6 1D)30 targeting ER AF 2 or morespecifically the surface charge clamp. High throughput screening (HTS) of large, 10,000strong, compound libraries in combination with follow up optimization studies (SAR31/X rayscreen32) has provided an alternative approach to identifying inhibitors of AF 2 binding ofER (Figure 6 1E),33,31 AR,32 but also AR binding function 3 (BF 3), which is an allostericbinding site in proximity to AF 2 (Figure 6 1F).32

Figure 6 1 | Nuclear receptor coactivator binding inhibitorsSmall lipophilic drug like molecules targeting activation function 2 (AF 2) or binding function 3(BF 3) of NRs: (A,B) peptidomimetic pyrimidines targeting ER and AR, respectively;28(C) pyridylpyridone binding to ER;29 (D) biphenyl proteomimetic interacting with ER;30(E) quinazolinone derivatives;31 (F) pyrazolo pyrimidine targeting AR.32

NPs are also interesting as potential lead structures for the development of novelantagonists of the retinoid X receptors (RXR , NR2B1; RXR , NR2B2). RXR is a type II NR,which on binding of the putative endogenous ligand, 9 cis retinoic acid, can eitherhomodimerize or form heterodimer pairs with other type II NRs such as the peroxisomeproliferator activated receptors (PPARs, NR1CX) and the vitamin D receptor (VDR,NR1I1).34–37 RXR plays a major role in the regulation of the gene networks connected to cellgrowth, cell differentiation, and cell death.37 Rexinoids, RXR selective ligands, are used in thetreatment of cutaneous T cell lymphoma and additional are in clinical trial for both breastand lung cancer.38–41 Furthermore, via modulation of pathways based on RXR heterodimerformation, including LXR and PPAR, such compounds could be useful in the treatment of

Chapter Six

126

cardiovascular and inflammatory diseases, including diabetes, obesity, metabolic syndrome,and atherosclerosis.39,42–47 In two recent independent studies the biaryl NP honokiol (Figure 62, 1) has been discovered as a suspected partial RXR agonist.48,49 From a drug discoveryperspective, this molecule is interesting due to its high molecular efficiency, its readyavailability and ease of synthesis. Furthermore, biaryls are privileged structures for drugdiscovery, since they have shown a high capacity to bind a range of different proteins.50

Figure 6 2 | Biaryl neolignan secondary metabolites isolated fromMagnolia species

Honokiol and its natural isomer magnolol (Figure 6 2, 2) are major components of thedried bark of theMagnolia (M.) officinalis andM. obovata, which for centuries has been used intraditional eastern medicine for the treatment of a variety of ailments, including indigestion,anxiety and depression.51 More recently, 1 and 2 have been isolated along with related biarylmetabolites, such as 4’ O methyl honokiol and 2’ O methyl honokiol (Figure 6 2, 3 and 4,respectively) and have since exhibited a range of beneficial biological effects, including theinduction of neurite growth, neuronal protection and anti angiogensesis.52–54 The morecomplex neolignan sesquiterpene analogues, clovanemagnolol and caryolanemagnolol(Figure 6 2, 5 and 6, respectively), also display intriguing neuronal properties,55 which hasstimulated recent synthetic efforts and subsequent phenotype studies.56–62

6.2. Initial screen of biaryl natural products isolated fromMagnolia species

First insights into the ability of RXR to interact with downstream proteins uponinteraction with Magnolia derived biaryl natural products were obtained via a microarrayassay for nuclear receptor coregulator interaction profiling. This assay measures the ability ofNRs to recruit a range of biologically relevant co activator peptides to the receptor surface(Figure 6 3).63 The small compound library featured alongside commercially available 1 and


127

2, the known natural products mono O methyl analogues of 1 (3 and 4, respectively) and themore complex neolignan sesquiterpenes 5 and 6 (Figure 6 2, compounds 3 6 weresynthesized by LGM, LHP, CVL, MS). 55

Figure 6 3 | Preliminary screen of biaryl natural products isolated from theMagnolia species.(A) Representation of the principle of the microarray based profiling assay. Details andimmobilized peptides sequences: 6.9 Experimental, Nuclear receptor coregulator interaction profiling,Table 6 2.63 (B) Initial nuclear receptor coregulator interaction profiling at 200 M compoundconcentration identifies three different functionalities for biaryl compounds from Magnolia:honokiol (1) shows full inhibition of basal RXR activity, magnolol (2) and 4’ O methyl honokiol(3) show either partial activation or did not reach their maximum, and the others (4 6) show noeffect. As positive control: synthetic RXR agonist LG100268 10.64,65

Each of the NPs showed a unique NR coregulator interaction profile. While 2 showedenhanced overall binding compared to a DMSO control, 1 in comparison showed nearly totalsuppression of all interactions (Figure 6 3, 1 and 2, respectively). The neolignansesquiterpene analogues 5 and 6, however were found not to influence the RXR coregulatorinteraction. Interestingly, methylation of the 4’ hydroxyl group of 1 (resulting in 3)corresponded with an enhancement of coregulator binding, similar to 2. Methylation of the

N

HOO

10

HO

OH

1

B

A

Chapter Six

128

second 2’ hydroxyl group (resulting in 4), however, led to a compound silent on coactivatorbinding (Figure 6 3, 3 and 4, respectively).

6.3. Concentration dependent effects of honokiol with and without agonist

The suppressive function of 1 on RXR cofactor binding was subsequently tested atdifferent concentrations (0.01 100 M). While lower concentrations between 0.01 M and0.1 M showed only low or no inhibition of the RXR coregulator interaction, higherconcentrations, however, lowered the binding affinity significantly in a dose dependentmanner (Figure 6 4A). The introduction of the potent RXR agonist LG100268 (10)64 at 100 nMdid not prevent 1 from binding to RXR . Although receptor activity was initially enhancedby 10 resulting in a positive effect (>intrinsic receptor activity), this was reversed due to thefunction of 1 to block cofactor interaction at higher concentrations. It is important to mentionat this stage that the final negative effect of 1 the ability of 1 to inhibit coactivator bindingwas not affected by the presence of 10 (Figure 6 4B). This indicates that 1 and 10 are notcompeting for a binding site on the receptor surface.

Figure 6 4 | Concentration dependent effect of honokiol on the RXR coregulator interactionInfluence of increasing concentrations of 1 on RXR coregulator interaction . (A) Without and (B)in competition with 100 nM agonist 10.

6.4. Insights into the honokiol RXR interaction

A more indepth characterization of the effect of RXR on the concentration dependenteffects of 1 was performed using a polarization assay. Agonist activated RXR LBD (3.2 Mof 10) bound to the fluorescein labeled peptide D22 indentified from phage displayscreening against ER 66 afforded good binding affinity to RXR. Addition of increasingconcentrations of 1 to this agonist activated receptor resulted in the disruption of the RXR –peptide interaction at high concentrations of 1. Importantly, this inhibitory function of 1 wasinsensitive to the concentration of agonist 10 (Figure 6 5A), supporting again the notion ofdifferent receptor interaction sites. In a separate experiment, increasing concentrations of 10

A B


129

produced a gradual activation of RXR , which leveled off at saturating concentrations of theagonist. The addition of fixed concentrations of 1 to the receptor resulted in a lowering of themaximal effect achievable by 10. In contrast, the values for the half maximal effectiveconcentration remained constant upon addition of 1 (EC50, Figure 6 5B). Taken together theseresults thus identified 1 as a non competitive antagonist most probably a coactivatorbinding inhibitor (CBI) which does not target the ligand binding pocket of RXR .

Figure 6 5 | Polarization assay of RXR with honokiol and agonist 10(A) Increasing concentrations of 1 led to a decrease in the ability of RXR activated by 10 – torecruit D22, which binds to the coactivator binding surface. Higher concentrations of 10 did notchange the inhibitory property of 1. (B) Increasing concentrations of 10 led to activation of RXR ,which resulted in an increase in polarization. The presence of 1 resulted in a lowering of themaximal effect, but did not change the half maximal effective concentration (EC50).

6.5. Studies on the pharmacological profile of honokiol

Getting insights in the biological relevance of honokiol as possible CBI both cellular andselectivity studies were performed. In a mammalian two hybrid assay format 1 showedagonistic properties at a concentration of between 10 25 M. This is in line with other studiesthat reported a partial agonistic effect of 1 for RXR.48,49 A further increase in concetration,however, led to a reduced luciferase activity. This reduced signal at 50 M could also beobserved for the same experiment in the presence of 10 nM of 10. However, cytotoxicity of 1rather than CBI activity might be an explanation for this decrease in signal.67

The selectivity of 1 for RXR was evaluated on chip performing a NR coregulatorinteraction profiling for the LBDs of RXR , RXR , ER and AR in presence of an respectiveagonist. Receptor cofactor binding was evaluated in the presence of 100 nM agonist alone(RXR: 10, ER: estradiol, AR: dihydrotestosterone) and in combination with 100 M of 1 as anintended non competitive antagonist. Both RXR and RXR showed a significant decrease intheir ability to recruit cofactors upon addition of 1 compared to the receptor treated with theagonist alone. However, neither ER nor AR were effected by 1 (Figure 6 6). Honokiol, 1,

A B

Chapter Six

130

thus has exclusive inhibitory properties for RXR at the tested concnetration, but does notdiscriminate between different RXR subtypes.

Figure 6 6 | Biological relevance of honokiol (cellular activity/selectivity)(A) In a mammalian two hybrid assay the influence of 1 on RXR activity was tested with andwithout 10 M of agonist 10. (B)The ability to recruit cofactors in the presence of an agonist isblocked by 1 (A) in the case of RXR subtypes and , (C) but not in the case of ER or AR.Compounds: honokiol (1), RXR agonist LG100268 (10), ER agonist estradiol (E2) and AR agonistdihydrotestosterone (DHT).

A

C E2 DHT

B

O

OH

HH

H

DHT


131

6.6. Modification of honokiol towards exclusive targeting of the RXR AF 2

Jung et al. and Kotani et al. discovered partial agonistic effects at low concentrations of 1that is based on binding to the LBP.48,49 Our study revealed additional non competitiveantagonistic effects of 1 at higher concentrations which requires binding to an alternativebinding site, possibly the coactivator binding surface (activation function 2, AF 2). Todiscriminate this dual effect in favor of AF 2 binding, chemical modifications wereintroduced into 1 to mimic the LXXLL motif of coactivator peptides. The two allyl groupspresent in 1 were replaced by propyl , epoxymethyl , or cyclopropylmethyl side chains asleucine side chain mimics with varying degrees of lipophilicity and steric bulk (Figure 6 7, 7,8, 11). Furthermore, allyl side chains were conveniently installed in a speculative mannerortho to the hydroxyl group on the aromatic rings to extend lipophilicity in this region of themolecule in search of improved binding at the lipophilic RXR surface (Figure 6 7, 9, 12).

Figure 6 7 | Design strategy to enhance binding properties at the alternative binding siteModification of 1 to optimize leucine mimic: either by modification of two allyl groups by propyl(7), 2,3 epoxypropyl (8), and cyclopropylmethyl substitution (11) or through the introduction ofone (9) or two (12) additional allyl groups.

Initial screening identified 7, 8, and 11 to have the best inhibitory properties (data notshown). In a second screen in the presence of 100 M agonist 10, both the epoxypropylsubstitution (8) and the introduction of one additional allyl group (11) corresponded with asustained ability to inhibit RXR coactivator recruitment. However, the activity was slightlyreduced compared to 1. Adding more flexibility through substitution of the two allyl groupswith propyl side chains (7) even enhanced the inhibitory properties (Figure 6 8).

Chapter Six

132

Figure 6 8 | Honokiol analogues exclusively targeting the coactivator binding surfaceProfiling the RXR coregulator interaction in the presence of 100 M of agonist 10 and 100 M of7 9 and 11 12.

6.7. Discussion

Profiling the retinoid X receptor (RXR ) coregulator interaction against a set of biarylnatural products (NPs) resulted in unique coregulator binding profiles (Figure 6 2, 1 6). At200 M treatment with 2 led to an activation of the receptor regarding cofactor binding,while treatment with 1 on the other hand induced a loss in RXR cofactor recruitment ability.Interestingly, modification of 1 to give mono O methylated analogues 3 resulted in partialactivation of coregulator recruitment (relative to 2), while the isomeric mono O methylatedanalogue 4 was effectively inactive. This would appear to suggest that at 200 M, 1 functionsas an inverse agonist or coactivator binding inhibitor (CBI), inhibiting basal coregulatorbinding ability of RXR, while analogues 2 and 3 induce coactivator recruitment in a partialmanner (partial agonist). It is important to emphasize here how the small differencesbetween the isomers 1 and 2 produces opposite effects. And the minor change from 1 to 3,namely the introduction of a methyl group, seems to switch activity from inhibition to partialagonist behaviour. The effects of clovanemagnolol (5) and caryolanemagnolol (6) in thisassay were not significant, though the possibility that these two molecules bind to the LBP asantagonists and block co activator could not be ruled out at this stage. At low concentrationsof 1 no noticeable change in the coactivator profile was observed (Figure 6 4, 10 100 nM),which would indicate that inhibition of basal coactivator binding does not occur within thisconcentration range. 1 might still bind to the LBP. However, no activation of coactivatorbinding could be detected in the microarray format. At higher concentrations (Figure 6 4,1 100 M), a dose dependent decrease in the fluorescence signal was observed similar towhat had been observed during the initial 200 M screen. In the presence of 100 nM of aknown RXR agonist 10, a similar inhibitory pattern was observed. Furthermore, competitivefluorescence polarization studies of 1 revealed that coactivator inhibition was insensitive to


133

changes in the concentration of the agonist 10 (Figure 6 5). This is in line with the profilingdata, and suggests that 1 inhibits coactivator binding via an alternative mechanism and doesnot function as an inverse agonist targeting the RXR LBP. Investigating the exclusiveness ofRXR binding, the effect of 1 on four different NRs was tested. In this case, the proteins wereincubated in the presence of 100 M of 1 and 100 nM of a known full agonist – E2 (ER ),DHT (AR), 10 (RXR and RXR ) – and the effect on coregulator binding monitored in themicroarray format. While only minor changes were observed in the case of type I NRs ERand AR, a clear decrease in signal was measured for both RXR and RXR (Figure 6 6). Thissuggests that 1 behaves as an alternative NR co activator inhibitor with selectivity for RXR(albeit without sub type selectivity), and thus represents a new molecular structure for thepotential development of RXR selective inhibitors.The results obtained via mammalian two hybrid assays (M2H, Figure 6 6A) are in line with

data from TR FRET and reporter gene assays reported in the literature.48,49 These experimentsprovide evidence that 1 binds to the RXR LBP at lower, micromolar concentrations(10 30 M). However, as suggested in the M2H assay and confirmed during NR coragulatorprofiling, 1 is also capable of displaying an inhibitory effect. Thus, 1 most likely binds theLBP at lower concentrations, leading to weak co activator recruitment, while at higherconcentrations, a second molecule 1 binds at an allosteric site away from the LBP resultingin a concentration dependent inhibition of coactivator binding (CBI). The ability of NRs tobind the same molecule at two distinct sites on the protein has already been reported in thecase of 4 hydroxytamoxifen and its ability to bind LBP and AF 2 of ER, but would representa novel finding for RXR.68,69 Furthermore, the partial agonist properties measured for 2(Figure 6 3) are concordant with structural studies showing 2 co crystallized with RXR ,70

and suggest that 1might be equally suited for binding at the RXR LBP as 2.The use of co activator binding inhibitors (CBIs) has emerged as a highly promising

alternative strategy to classical LBP agonism/antagonism, though RXR selective CBIs havenot been reported thus far. CBIs target the NR coactivator interaction by mimicking theleucine side chains of the LXXLL peptide motif, which is conserved in several NRcoactivators (peptidomimetic). This has typically been achieved through the incorporation oflipophilic side chains to a rigid, heterocyclic scaffold, leading to some notable successes inthe case of ER and AR.28 Despite this, though, novel CBIs are still needed in order to expandon the current repertoire; especially compounds with selectivity for specific NR proteinclasses.71 Interestingly, Katzenellenbogen et al. recently disclosed the development ofbiphenyl derived inhibitors of the ER co activator interaction, with specific designs for theAF 2.30 Central to the design of this compound class was the incorporation of lipophilicgroups that mimic the leucine side chains of the signature LXXLL motif of the coactivatorrecognition sequence. In light of the structural similarity between 1 and known NR CBI’s(Figure 6 1, e.g. D) it could therefore be reasonably hypothesized that the two allyl groups

Chapter Six

134

featured in 1 might also mimic the same leucine side chains. The relative orientation of theallyl groups is important in this case, as is the ability of the biaryl bond to rotate enabling anout of plane configuration of the allyl substituents. Since 1 is also based on a biaryl corestructure it meets all these requirements important for NR AF 2 targeting.29,30

Investigations into 1 focusing on the optimization of the allylic side chains shouldstrengthen the hypothesis that 1 binds to the AF 2 as CBI and not an alternative site on thereceptor surface.71 Thus additional allyl groups were introduced ortho to the hydroxyl groupsof 1 (Figure 6 7A, 9 and 12, respectively) or the two existing allylic groups were substitutedwith propyl (7), 2,3 epoxypropyl (8), or cyclopropylmethyl (11) to produce a small libraryvarying in lipophilicity and possibly featuring higher affinity for RXR AF 2. This approach ofsidechain optimization in search of improved bidning to AF 2 has been successfullyperformed for other NRs.72,73 While 9 showed inhibitory activity similar to 1, the analoguecontaining four allylic subsituents (12) showed a significantly reduced activity; althoughactivity was not entirely abolished. One explanation could be that the introduction of thefourth allyl group on 1 (or 9) leads to a steric hindrance and a consequent limited access ofthe hydroxyl group, which that play an important role in RXR binding, for example througha hydrogen bond interaction with one of the charge clamp amino acid residues.30

Surprisingly, 11 was one of the weakest binders to emerge from the small library. In contrast,8 showed a good ability to block RXR cofactor interaction, similar to 1, and analogue 7 evenexceeded the inhibitory capabilities of 1. This preliminary data suggests that 1 one is indeedtargeting the AF 2 of RXR. However, further evaluation of these compounds required.Structure activity relationship studies of this natural product 1 with high molecularefficiency reveal that minor modifications such as the introduction of a methyl group(1 to 3) or a repositioning of the hydroxyl group (1 to 2) can lead to a switch between partialagonism and coactivator binding inhibition. Furthermore, first trials in the modulation of theinhibitory function through modifications at the side chain positions show promising results.


135

6.8. Conclusion

Initial microarray based RXR coregulator interaction profiling of a focused natural productcompound library of biaryl molecules isolated from Magnolia species has resulted in thediscovery of three classes of molecule with distinct modes of RXR binding: no agonistic effectwith possible antagonistic properties (4 6), partial agonism (2 and 3, respectively), andinhibitory function as an inverse agonist or coactivator binding inhibitor (1). Furthercharacterization in polarization assay suggested that 1 does not target RXR via the LBP, butvia an alternative binding site. Subsequent diversification of 1 with the goal to optimize theleucine mimics known to be important for NR AF 2 binding revealed possible gradualmodulation through modifications to the allylic side chains (7 9,11,12).Thus the compounds isolated from Magnolia bark can either directly or via close structural

analogues function as (partial) agonist and antagonists targeting both the LBP or as CBIsblocking the AF 2 coactiavtor interaction. Taken together, these results provide furtherconvincing evidence of the power of natural products as a source of novel structures andnovel activities for NR based drug discovery.

Chapter Six

136

6.9. Experimental

GeneralChemical synthesis of compound library performed by Dr. Lech Gustav Milroy, Leslie in Het Panhuis,Chan Vinh Lam, Nicky Hoek and Marcel Scheepstra.

If not stated otherwise the following contents, descriptions, protocols, and distributer are valid:chemicals were ordered from SIGMA ALDRICH. Proteins were handled at 4 °C. Standard culturingmedium is lysogeny broth (LB) medium (10 g peptone, 10 g NaCl, 5 g yeast, 1 l water, autoclaved).



construct vector tag protein originAR LBD pGEX KG GST hAR LBD664 919 BAYER HEALTHCARE PHARMACEUTICALS



Plasmids with the desired protein construct (Table 6 1) were transformed into E. coli BL21 cells.Emerging colonies were cultured over night in 25 ml LB medium with 100 g/ml ampicillin at 37 °C.This preculture was added to 2 l phosphate buffered, content rich terrific broth (TB) medium (12 gpeptone, 24 g yeast, 4 ml glycerol, 0.17 M KH2PO4, 0.72 M K2HPO4, 1 l water, autoclaved) withampicillin and incubated at 37 °C until an optical density at 600 nm (OD600) of 0.6 1.0. Addition of100 M Isopropyl D 1 thiogalactopyranoside (IPTG) induced expression of the protein of interest.For stabilization 10 M of ligand (ER, estradiol | AR, dihydrotestosteron | RXR, agonist 10) wasadded. After incubation for 20 h at 15 °C the cells were harvested by centrifugation (15 min at 7,000rpm) and the resulting pellet was stored at 80 °C until further usage.If not stated otherwise, the protein purification was performed at 4 °C. The pellet was resuspended inlysis buffer (1x PBS tablet, CALBIOCHEM, 370 mM NaCl, 10% (v/v) glycerol, 0.1 mM TCEP, 1 mM PMSF,1 g/ml DNAseI, 10 M ligand, pH 8.0, His6 tag: additional 40 mM imidazole) and lyzed with theEmulsi Flex C3 homogenizer (2 passes of 150,000 kPa, AVESTIN INC.). After another centrifugation step(40 min, 20,000 rpm) the supernatant was provided on the respective affinity chromatography column(GST tag: Protino GST/4B, 1 ml, MACHEREY NAGEL | His6 tag: His Bind Resin, 3 ml, NOVAGEN),washed with wash buffer (GST tag: 1x PBS tablet, CALBIOCHEM, pH 7.3 | His6 tag: 1x PBS tablet,CALBIOCHEM, 370 mMNaCl, 40 mM imidazole, 10% (v/v) glycerol, 0.1 mM TCEP, 10 M ligand,pH 8.0) and eluted with competing molecules (GST tag: 50 mM Tris, 10 mM glutathione | His6 tag:wash buffer + 100 mM / 250 mM / 500 mM imidazole). Fractions were analyzed with SDS PAGE andoptional with LC MS.

Nuclear receptor coregulator interaction profiling63,75

Assay was prepared on ice in three different mastermixes with TR FRET coregulator buffer(INVITROGEN) with 1 M DTT (RXR: buffer G, ER: buffer E, AR: buffer A): (1) 4X anibody mix,containing 100 nM antibody (GST: anti GST Alexa488, INVITROGEN; His: PentaHis Alexa488, QIAGEN)(2) 4X protein mix, containing 10 or 40 nM protein (GST RXR: 40 nM; His6 ER, GST AR: 10 nM), and2X ligand mix, containing double concentration of final compound concentration in 2% (v/v)DMSO.Combination of them is resulting in 25 l solution including 25 nM antibody and 2.5 nM or 10 nM


137

protein and the compound in desired concentration (10 nM – 200 M). All assays were performed inPamStation® 4 controlled by EvolveHT software (PAMGENE INTERNATIONAL B. V., S HERTOGENBOSCH,THE NETHERLANDS) at 28 °C. The Nuclear Receptor PamChip® Arrays (PAMGENE) contain 53 peptides(Table 6 2) immobilized on a 3D cylindric structured material with branched pores with 100 m indiameter and 60 m in length. This resulted in a approximate 500 times enlarged surface area that isdirectly correlated to the read out signal. 20 initial cycles with 25 l blocking buffer (1% BSA, 0.01%Tween 20 in Tris buffered saline) are followed by 25 l of assay mixture. In the time period of 40minutes the solution containing protein, fluorescent antibody and ligand were pumped up and down(81 times, 2 cycles/min) through the porous membrane with the immobilized peptides (all 53 inparallel). The fluorescent signal was detected every 20 cycles by a charge coupled device camerabased optical system interacted in the instrument. Image analysis was performed by Bionavigatorsoftware (PAMGENE), including automated spot finding and quantitation of signal minus backgroundvalues (100 ms).

Mammalian Two Hybrid assayAll Mammalian Two Hybrid assays (M2H) were performed in human osteosarcoma cells (U2 OS76)seeding 30,000 cells/well in a 24 well plate (BD Biosciences). Plasmids are either from MammalianTwo Hybrid Assay Kit (AGILENT) or cloned by Dr. Ingrid J. de Vries van Leeuwen. Culturingconditions were 37 °C and 5% CO2. Cells were transfected with following amounts of DNA/well usinglinear polyethylenimines (PEI, POLYSCIENCES): (1) 40 ng of peptide construct pCMV LXXLL containingHPLLMRLLLSP peptide discovered to bind effectively RXR fused to DNA binding domain of theGAL4, (2) 40 ng of the target construct pCMV AD RXR , containing RXR fused to the transcriptionalactivation domain (AD) of the mouse protein NF B, (3) 200 ng of the reporter plasmid pFR Luc withthe synthetic promoter with five tandem repeats of the GAL4 binding sites that control expression ofthe Photinus pyralis firefly luciferase, and (4) 3.2 ng of control plasmid Renilla Luc, expressingindependently Renilla reniformis luciferase. After 16 h incubation cells were treated with 10 nM agonistLG100268 with varying concentrations of honokiol (10 nM – 50 M). After additional 24 h incubationthe interaction was determined with a Dual Luciferase Reporter Assay (PROMEGA), according tomanufacturer’s instruction. The luminescent intensities were recorded on a Synergy HT Multi ModeMicroplate Reader (BIOTEK). The interaction dependent Firefly luciferase signal was first normalizedover the independent Renilla luciferase signal and subsequently normalized over the ratio of untreatedcells. All cellular studies were performed by Dr. Ingrid J. de Vries van Leeuwen.

Nuclear receptor coregulator interaction profiling63,75

Assay was prepared on ice in three different mastermixes with TR FRET coregulator buffer(INVITROGEN) with 1 M DTT (RXR: buffer G, ER: buffer E, AR: buffer A): (1) 4X anibody mix,containing 100 nM antibody (GST: anti GST Alexa488, INVITROGEN; His: PentaHis Alexa488, QIAGEN)(2) 4X protein mix, containing 10 or 40 nM protein (GST RXR: 40 nM; His6 ER, GST AR: 10 nM), and2X ligand mix, containing double concentration of final compound concentration in 2% (v/v)DMSO.Combination of them is resulting in 25 l solution including 25 nM antibody and 2.5 nM or 10 nMprotein and the compound in desired concentration (10 nM – 200 M). All assays were performed inPamStation® 4 controlled by EvolveHT software (PAMGENE INTERNATIONAL B. V., S HERTOGENBOSCH,THE NETHERLANDS) at 28 °C. The Nuclear Receptor PamChip® Arrays (PAMGENE) contain 53 peptides(Table 6 2) immobilized on a 3D cylindric structured material with branched pores with 100 m indiameter and 60 m in length. This resulted in a approximate 500 times enlarged surface area that isdirectly correlated to the read out signal. 20 initial cycles with 25 l blocking buffer (1% BSA, 0.01%

Chapter Six

138

Tween 20 in Tris buffered saline) are followed by 25 l of assay mixture. In the time period of 40minutes the solution containing protein, fluorescent antibody and ligand were pumped up and down(81 times, 2 cycles/min) through the porous membrane with the immobilized peptides (all 53 inparallel). The fluorescent signal was detected every 20 cycles by a charge coupled device camerabased optical system interacted in the instrument. Image analysis was performed by Bionavigatorsoftware (PAMGENE), including automated spot finding and quantitation of signal minus backgroundvalues (100 ms).

Fluorescence Polarization assay77,78

Both fluorescence polarization (FP) experiments were measured in black 384 well plates (PERKINELMER 384 F) on a Safire2 (TECAN) plate reader. The final volume of one well was 30 l containingconstant concentrations of ligand (5 M) and Fluorescein labeled peptide (D22,LPYEGSLLLKLLRAPVEEV, INVITROGEN, phage display hit ER LBD, good binding propertiesRXR 66). The protein was sequentially diluted (24 steps) in TR FRET coregulator buffer E(INVITROGEN). Briefly, in 96 well plates dilution series were prepared in 55 l and filled up with to110 l with a 2X master mix solution containing all non varying components. Finally three times 30 lwas transferred to 384 well plates, centrifuged (1,000 rpm, 2 min) and incubated for 1 h at 4 °C. Plateswere measured 50 times at 30 °C (excitation 470 nm, emission 519 nm). Polarization (in [mP]) wasplotted against the concentration of either the protein, whereby each data point represents anaverage of 3 experiments.The dissociation constant Kd of the protein peptide complex (FP) was calculated using equation 1:



139

Table 6 2 | Coregulator sequences immobilized on microarray

position protein sequence motif uniprot1 CBPLxxLL2067 SVQPPRSISPSALQDLLRTLKSP LxxLL2067 Q92793 2 CBPLxxLL358_C367S TADPEKRKLIQQQLVLLLHAHKSQ LxxLL358 Q92793 3 CBPLxxLL70 GNLVPDAASKHKQLSELLRGGSGS LxxLL70 Q92793 4 DAX1LxxLL146 GEDHPRQGSILYSLLTSSKQTHVA LxxLL146 P51843 5 DAX1LxxML13 MAGENHQWQGSILYNMLMSAKQT LxxML13 P51843 6 DAX1LxxML80 FSGKDHPRQGSILYSMLTSAKQT LxxML80 P51843 7 CBPLxxLL358 TADPEKRKLIQQQLVLLLHAHKCQ LxxLL358 Q92793 8 P300LxxLL2051 SPLKPGTVSQQALQNLLRTLRSP LxxLL2051 Q09472 9 P300LxxLL81 GMVQDAASKHKQLSELLRSGSSP LxxLL81 Q09472 10 IKBBLxxLL289 PLGSAMLRPNPILARLLRAHGAP LxxLL289 Q15653 11 IKBBLxxLL74 LHLAVIHQHEPFLDFLLGFSAGT LxxLL74 Q15653 12 TRIP8LxxLL2066 PLVSQNNEQGSTLRDLLTTTAGK LxxLL2066 Q15652 13 TRIP5LxxLL844 QWVSSLNEREQELHNLLEVVSQS LxxLL844 P52732 14 SRC1LxxLL1435 TSGPQTPQAQQKSLLQQLLTE LxxLL1435 Q15788 15 SRC1LxxLL633 SDGDSKYSQTSHKLVQLLTTTAEQ LxxLL633 Q15788 16 SRC1LxxLL690 PSSHSSLTERHKILHRLLQEGSPS LxxLL690 Q15788 17 SRC1LxxLL749 ASKKKESKDHQLLRYLLDKDEKD LxxLL749 Q15788 18 TIF2LxxLL641 GQSRLHDSKGQTKLLQLLTTKSDQ LxxLL641 Q15596 19 TIF2LxxLL690 STHGTSLKEKHKILHRLLQDSSSP LxxLL690 Q15596 20 TIF2LxxLL745 EPVSPKKKENALLRYLLDKDDTK LxxLL745 Q15596 21 PGC1 LxxLL144 PPQEAEEPSLLKKLLLAPANT LxxLL144 Q9UBK2 22 TIF2LxxLL878 SQSTFNNPRPGQLGRLLPNQNLP LxxLL878 Q15596 23 SRC3LxxLL113 KKKGQGVIDKDSLGPLLLQALDG LxxLL113 Q9Y6Q9 24 SRC3LxxLL621 QRGPLESKGHKKLLQLLTCSSDD LxxLL621 Q9Y6Q9 25 SRC3LxxLL621_C627S QRGPLESKGHKKLLQLLTSSSDD LxxLL621 Q9Y6Q9 26 SRC3LxxLL685 MHGSLLQEKHRILHKLLQNGNSP LxxLL685 Q9Y6Q9 27 mSRC3LxxLL1041 HGSQNRPLLRNSLDDLLGPPSNA LxxLL1041 O09000 28 PGC1 LxxLL156 PAPEVDELSLLQKLLLATSYP LxxLL156 Q86YN6 29 PRIPLxxLL1491 LVSPAMREAPTSLSQLLDNSGAP LxxLL1491 Q14686 30 PRIPLxxLL887 PVNKDVTLTSPLLVNLLQSDISA LxxLL887 Q14686 31 NCORLxxHI2051 MGQVPRTHRLITLADHICQIITQ LxxHI2051 O75376 32 NCORLxxHI2051_C2056S MGQVPRTHRLITLADHISQIITQ LxxHI2051 O75376 33 NCORLxxII2263 GHSFADPASNLGLEDIIRKALMG LxxII2263 O75376 34 SMRTLxxHI2135 APGVKGHQRVVTLAQHISEVITQ LxxHI2135 Q9Y618 35 PGC1 LxxLL343 AEFSILRELLAQDVLCDVSKP LxxLL343 Q86YN6 36 SMRTLxxII2342 QAVQEHASTNMGLEAIIRKALMG LxxII2342 Q9Y618 37 RIP140LxxLL133(P124R) DSVRKGKQDSTLLASLLQSFSSR LxxLL133 P48552 38 RIP140LxxLL185 KDLRCYGVASSHLKTLLKKSKVK LxxLL185 P48552 39 RIP140LxxLL185(C177S) KDLRSYGVASSHLKTLLKKSKVK LxxLL185 P48552 40 RIP140LxxLL380 RNNIKQAANNSLLLHLLKSQTIP LxxLL380 P48552 41 RIP140LxxLL501 KNSKLNSHQKVTLLQLLLGHKNE LxxLL501 P48552 42 RIP140LxxLL713 SEIENLLERRTVLQLLLGNPTKG LxxLL713 P48552 43 RIP140LxxLL819 PVSPQDFSFSKNGLLSRLLRQNQDSYL LxxLL819 P48552 44 RIP140LxxLL936 RSWARESKSFNVLKQLLLSENCV LxxLL936 P48552 45 RIP140LxxLL936(C945S) RSWARESKSFNVLKQLLLSENSV LxxLL936 P48552 46 GCN5LxxLL190 EEDADTKQVYFYLFKLLRKSILQ LxxLL190 Q92831 47 PBPLxxLL604 HGEDFSKVSQNPILTSLLQITGNG LxxLL604 Q15648 48 PBPLxxLL645 VSSMAGNTKNHPMLMNLLKDNPAQ LxxLL645 Q15648 49 PGC1 LxxLL144 DGTPPPQEAEEPSLLKKLLLAPANTQ LxxLL144 Q9UBK2 50 SHPLxxIL118 TFEVAEAPVPSILKKILLEEPSS LxxIL118 Q15466 51 SHPLxxLL21(C9S/C11S) SPSQGAASRPAILYALLSSSLKA LxxLL21 Q15466 52 TRIP4LxxLL161(C171S) FVNLYTRERQDRLAVLLPGRHPS LxxLL161 Q15650 53 TRIP3LxxLL101 LQNLKNLGESATLRSLLLNPHLR LxxLL101 Q15649

Chapter Six

140

6.10. References

1. I. J. McEwan, in The Nuclear Receptor Superfamily, ed. I. J. McEwan, Humana Press, Totowa, NJ,USA, 2009, 505, 3–18.

2. P. Huang, V. Chandra, and F. Rastinejad, Annu Rev Physiol, 2010, 72, 247–272.3. I. G. Schulman, Adv Drug Deliv Rev, 2010, 62, 1307–1315.4. D. J. Newman and G. M. Cragg, J Nat Prod, 2012, 75, 311–335.5. F. Lovering, J. Bikker, and C. Humblet, J Med Chem, 2009, 52, 6752–6756.6. R. Breinbauer, I. R. Vetter, and H. Waldmann, Angew Chem Int Ed, 2002, 41, 2879–2890.7. M. Kaiser, S. Wetzel, K. Kumar, and H. Waldmann, Cell Mol Life Sci, 2008, 65, 1186–1201.8. K. Kumar and H. Waldmann, Angew Chem Int Ed, 2009, 48, 3224–3242.9. S. C. K. Sukuru, J. L. Jenkins, R. E. J. Beckwith, J. Scheiber, A. Bender, D. Mikhailov, J. W. Davies,

and M. Glick, J Biomol Screen, 2009, 14, 690–699.10. D. H. Drewry and R. Macarron, Curr Opin Chem Biol, 2010, 14, 289–298.11. S. Wetzel, R. S. Bon, K. Kumar, and H. Waldmann, Angew Chem Int Ed, 2011, 50, 10800–10826.12. E. D. Walter, J Am Chem Soc, 1941, 63, 3273–3276.13. G. G. J. M. Kuiper, J. G. Lemmen, B. Carlsson, J. C. Corton, S. H. Safe, P. T. van der Saag, B. van

der Burg, and J. Å. Gustafsson, Endocrinology, 1998, 139, 4252–4263.14. N. L. Urizar, A. B. Liverman, D. T. Dodds, F. V. Silva, P. Ordentlich, Y. Yan, F. J. Gonzalez, R. A.

Heyman, D. J. Mangelsdorf, and D. D. Moore, Science, 2002, 296, 1703–1706.15. J. Wu, C. Xia, J. Meier, S. Li, X. Hu, and D. S. Lala,Mol Endocrinol, 2002, 16, 1590–1597.16. A. H. Payne and D. B. Hales, Endocr Rev, 2004, 25, 947–970.17. G. V. Satyavati, Indian J Med Res, 1988, 87, 327–335.18. V. Singh, S. Kaul, R. Chander, and N. K. Kapoor, Pharmacol Res, 1990, 22, 37–44.19. S. De Marino, R. Ummarino, M. V. D’Auria, M. G. Chini, G. Bifulco, C. D’Amore, B. Renga, A.

Mencarelli, S. Petek, S. Fiorucci, and A. Zampella, Steroids, 2012, 77, 484–495.20. I. A. Siddiqui, M. Asim, B. B. Hafeez, V. M. Adhami, R. S. Tarapore, and H. Mukhtar, FASEB J,

2011, 25, 1198–1207.21. D. Roell and A. Baniahmad,Mol Cell Endocrinol, 2011, 332, 1–8.22. H. R. Siddique, S. K. Mishra, R. J. Karnes, and M. Saleem, Clin Cancer Res, 2011, 17, 5379–5391.23. L. G. Meimetis, D. E. Williams, N. R. Mawji, C. A. Banuelos, A. A. Lal, J. J. Park, A. H. Tien, J. G.

Fernandez, N. J. de Voogd, M. D. Sadar, and R. J. Andersen, J Med Chem, 2012, 55, 503–514.24. I. J. McEwan, Biochem J, 2012, 446, e1–e3.25. A. L. Rodriguez, A. Tamrazi, M. L. Collins, and J. A. Katzenellenbogen, J Med Chem, 2004, 47, 600–

611.26. H. B. Zhou, M. L. Collins, J. R. Gunther, J. S. Comninos, and J. A. Katzenellenbogen, Bioorg Med

Chem Lett, 2007, 17, 4118–4122.27. A. A. Parent, J. R. Gunther, and J. A. Katzenellenbogen, J Med Chem, 2008, 51, 6512–6530.28. J. R. Gunther, A. A. Parent, and J. A. Katzenellenbogen, ACS Chem Biol, 2009, 4, 435–440.29. J. Becerril and A. D. Hamilton, Angew Chem Int Ed Engl, 2007, 46, 4471–4473.30. A. B. Williams, P. T. Weiser, R. N. Hanson, J. R. Gunther, and J. A. Katzenellenbogen, Org Lett,

2009, 11, 5370–5373.31. A. Sun, T. W. Moore, J. R. Gunther, M. S. Kim, E. Rhoden, Y. Du, H. Fu, J. P. Snyder, and J. A.

Katzenellenbogen, ChemMedChem, 2011, 6, 654–666.32. E. Estébanez Perpiñá, L. A. Arnold, A. A. Arnold, P. Nguyen, E. D. Rodrigues, E. Mar, R.

Bateman, P. Pallai, K. M. Shokat, J. D. Baxter, R. K. Guy, P. Webb, and R. J. Fletterick, Proc NatlAcad Sci USA, 2007, 104, 16074–16079.

33. A. L. LaFrate, J. R. Gunther, K. E. Carlson, and J. A. Katzenellenbogen, Bioorg Med Chem, 2008, 16,10075–10084.

34. D. J. Mangelsdorf, E. S. Ong, J. A. Dyck, and R. M. Evans, Nature, 1990, 345, 224–229.


141

35. D. J. Mangelsdorf, U. Borgmeyer, R. A. Heyman, J. Y. Zhou, E. S. Ong, A. E. Oro, A. Kakizuka, andR. M. Evans, Genes Dev, 1992, 6, 329–344.

36. D. J. Mangelsdorf, C. Thummel, M. Beato, P. Herrlich, G. Schütz, K. Umesono, B. Blumberg, P.Kastner, M. Mark, P. Chambon, and R. M. Evans, Cell, 1995, 83, 835–839.

37. P. Germain, P. Chambon, G. Eichele, R. M. Evans, M. A. Lazar, M. Leid, A. R. De Lera, R. Lotan,D. J. Mangelsdorf, and H. Gronemeyer, Pharmacol Rev, 2006, 58, 760–772.

38. M. F. Boehm, L. Zhang, L. Zhi, M. R. McClurg, E. Berger, M. Wagoner, D. E. Mais, C. M. Suto, andP. J. A. Davies, J Med Chem, 1995, 38, 3146–3155.

39. L. Altucci, M. D. Leibowitz, K. M. Ogilvie, A. R. de Lera, and H. Gronemeyer, Nat Rev DrugDiscov, 2007, 6, 793–810.

40. K. T. Liby, M. M. Yore, and M. B. Sporn, Nat Rev Cancer, 2007, 7, 357–369.41. P. Shankaranarayanan, A. Rossin, H. Khanwalkar, S. Alvarez, R. Alvarez, A. Jacobson, A.

Nebbioso, A. R. de Lera, L. Altucci, and H. Gronemeyer, Cancer Cell, 2009, 16, 220–231.42. T. Claudel, M. D. Leibowitz, C. Fiévet, A. Tailleux, B. Wagner, J. J. Repa, G. Torpier, J. M.

Lobaccaro, J. R. Paterniti, D. J. Mangelsdorf, R. A. Heyman, and J. Auwerx, Proc Natl Acad SciUSA, 2001, 98, 2610–2615.

43. P. Desreumaux, L. Dubuquoy, S. Nutten, M. Peuchmaur, W. Englaro, K. Schoonjans, B. Derijard,B. Desvergne, W. Wahli, P. Chambon, M. D. Leibowitz, J. F. Colombel, and J. Auwerx, J Exp Med,2001, 193, 827–838.

44. A. I. Shulman and D. J. Mangelsdorf, N Engl J Med, 2005, 353, 604–615.45. L. Michalik, J. Auwerx, J. P. Berger, V. K. Chatterjee, C. K. Glass, F. J. Gonzalez, P. A. Grimaldi, T.

Kadowaki, M. A. Lazar, S. O’Rahilly, C. N. A. Palmer, J. Plutzky, J. K. Reddy, B. M. Spiegelman, B.Staels, and W. Wahli, Pharmacol Rev, 2006, 58, 726–741.

46. D. D. Moore, S. Kato, W. Xie, D. J. Mangelsdorf, D. R. Schmidt, R. Xiao, and S. A. Kliewer,Pharmacol Rev, 2006, 58, 742–759.

47. O. Ziouzenkova and J. Plutzky, FEBS Letters, 2008, 582, 32–38.48. C. G. Jung, H. Horike, B. Y. Cha, K. O. Uhm, R. Yamauchi, T. Yamaguchi, T. Hosono, K. Iida, J. T.

Woo, and M. Michikawa, Biol Pharm Bull, 2010, 33, 1105–1111.49. H. Kotani, H. Tanabe, H. Mizukami, M. Makishima, and M. Inoue, J Nat Prod, 2010, 73, 1332–1336.50. P. J. Hajduk, M. Bures, J. Praestgaard, and S. W. Fesik, J Med Chem, 2000, 43, 3443–3447.51. N. Bushue and Y. J. Y. Wan, Adv Drug Deliv Rev, 2010, 62, 1285–1298.52. K. Watanabe, H. Y. Watanabe, Y. Goto, N. Yamamoto, and M. Yoshizaki, Jpn J Pharmacol, 1975, 25,

605–607.53. C. P. Hoi, Y. P. Ho, L. Baum, and A. H. L. Chow, Phytother Res, 2010, 24, 1538–1542.54. R. J. Leeman Neill, Q. Cai, S. C. Joyce, S. M. Thomas, N. E. Bhola, D. B. Neill, J. L. Arbiser, and J. R.

Grandis, Clin Cancer Res, 2010, 16, 2571–2579.55. Y. Fukuyama, Y. Otoshi, K. Miyoshi, K. Nakamura, M. Kodama, M. Nagasawa, T. Hasegawa, H.

Okazaki, and M. Sugawara, Tetrahedron, 1992, 48, 377–392.56. C. M. Chen and Y. C. Liu, Tetrahedron Lett, 2009, 50, 1151–1152.57. R. M. Denton, J. T. Scragg, A. M. Galofré, X. Gui, and W. Lewis, Tetrahedron, 2010, 66, 8029–8035.58. R. M. Denton, J. T. Scragg, and J. Saska, Tetrahedron Lett, 2011, 52, 2554–2556.59. S. Tripathi, M. H. Chan, and C. Chen, Bioorg Med Chem Lett, 2012, 22, 216–221.60. X. Cheng, N. L. Harzdorf, T. Shaw, and D. Siegel, Org Lett, 2010, 12, 1304–1307.61. Z. Khaing, D. Kang, A. M. Camelio, C. E. Schmidt, and D. Siegel, Bioorg Med Chem Lett, 2011, 21,

4808–4812.62. X. Cheng, N. Harzdorf, Z. Khaing, D. Kang, A. M. Camelio, T. Shaw, C. E. Schmidt, and D. Siegel,

Org Biomol Chem, 2012, 10, 383–393.63. A. Koppen, R. Houtman, D. Pijnenburg, E. H. Jeninga, R. Ruijtenbeek, and E. Kalkhoven, Mol Cell

Proteomics, 2009, 8, 2212–2226.

Chapter Six

142

64. R. Mukherjee, P. J. A. Davies, D. L. Crombie, E. D. Bischoff, R. M. Cesario, L. Jow, L. G. Hamann,M. F. Boehm, C. E. Mondon, A. M. Nadzan, J. R. Paterniti, and R. A. Heyman, Nature, 1997, 386,407–410.

65. M. F. Boehm, L. Zhang, L. Zhi, M. R. McClurg, E. Berger, M. Wagoner, D. E. Mais, C. M. Suto, andP. J. A. Davies, J Med Chem, 1995, 38, 3146–3155.

66. C. y Chang, J. D. Norris, H. Grøn, L. A. Paige, P. T. Hamilton, D. J. Kenan, D. Fowlkes, and D. P.McDonnell,Mol Cell Biol, 1999, 19, 8226–8239.

67. Y. J. Chen, C. L. Wu, J. F. Liu, Y. C. Fong, S. F. Hsu, T. M. Li, Y. C. Su, S. H. Liu, and C. H. Tang,Cancer Letters, 2010, 291, 20–30.

68. Y. Wang, N. Y. Chirgadze, S. L. Briggs, S. Khan, E. V. Jensen, and T. P. Burris, Proc Natl Acad SciUSA, 2006, 103, 9908–9911.

69. D. J. Kojetin, T. P. Burris, E. V. Jensen, and S. A. Khan, Endocr Relat Cancer, 2008, 15, 851–870.70. H. Zhang, X. Xu, L. Chen, J. Chen, L. Hu, H. Jiang, and X. Shen, PLoS ONE, 2011, 6, e28253.71. T. W. Moore, C. G. Mayne, and J. A. Katzenellenbogen,Mol Endocrinol, 2010, 24, 683–695.72. L. G. Milroy, L. Nieto, and L. Brunsveld, in Amino Acids, Peptides and Proteins, ed. E. Farkas and M.

Ryadnov, Royal Society of Chemistry, Cambridge, UK, 2012, 37, 238 272.73. P. T. Weiser, A. B. Williams, C. Y. Chang, D. P. McDonnell, and R. N. Hanson, Bioorg Med Chem

Lett, 2012, 22, 6587–6590.74. S. Folkertsma, P. I. van Noort, A. de Heer, P. Carati, R. Brandt, A. Visser, G. Vriend, and J. de

Vlieg,Mol Endocrinol, 2007, 21, 30–48.75. R. Houtman, R. de Leeuw, M. Rondaij, D. Melchers, D. Verwoerd, R. Ruijtenbeek, J. W. M.

Martens, J. Neefjes, and R. Michalides,Mol Cancer Ther, 2012, 11, 805–816.76. Y. Liang, D. F. Robinson, J. Dennig, G. Suske, and W. E. Fahl, J Biol Chem, 1996, 271, 11792–11797.77. P. Wu, M. Brasseur, and U. Schindler, Anal Biochem, 1997, 249, 29–36.78. F. Perrin, J Phys Radium, 1926, 7, 390–401.

Summary

III

Chemical Biology Approaches for Nuclear ReceptorsMolecular and Structural Insights

Nuclear receptors (NRs) are multi domain transcription factors that are controlled bysmall lipophilic molecules. Proteins of this protein superfamily feature domains with highsequence homology and domains which are comparatively diverse. Nevertheless the overallconformation of NRs is very similar. Small ligands interact with the ligand binding domain(LBD) of NRs via their buried ligand binding pocket (LBP). Dimerization of either twoidentical monomers or two different NRs is accomplished by structural rearrangements inthe receptor, mainly in the hinge region (HR) and the LBD. Subsequent interaction with NRresponse elements on DNA is controlled by the zinc fingers of the DNA binding domain(DBD) with partial involvement of the HR. Additional conformational change of the LBDleads to the establishment of the activation function 2 (AF 2). Together with activationfunction 1 (AF 1), located on the amino terminal domain (NTD), AF 2 controls coactivatorrecruitment and consequently transcriptional activation. Structural and functional studies ledto the finding that receptor specificity is connected to size and shape of the intermolecularinteraction interfaces, such as LBP and AF 2, but also to size and structure of varyingdomains, including NTD and C terminal F domain. The intrinsic structural flexibility of theAF 2 is one reason for the ability of NRs to respond differently to the presence of diversesmall molecules, including agonists, partial agonists, antagonists or reverse agonists. Thusboth transcriptional activation and repression via NRs are under functional control of liganddependent recruitment of coactivators and corepressors.Protein stability and activity are directly associated to posttranslational modifications

(PTMs) of NRs. These PTMs can have a considerable impact on conformational changesespecially in less structured regions and thus potentially modulate the functionality ofcertain protein domains and overall protein activity. In CHAPTER TWO acetylations of lysinesin the estrogen receptor (ER ) HR were investigated concerning their structural influenceon the mainly unstructured domain. Molecular dynamics (MD) simulation studies predictedstructural changes upon lysine acetylation, with double lysine acetylation exerting thegreatest effect in terms of their propensity to form helices. Structural analysis of 12 merpeptides synthesized by solid phase peptide synthesis by means of both CD and NMRstudies confirmed these results. The discovery of a possible presence of a transientpolyproline II (PPII) helix could be important in both the exploration of the molecularmechanism and specific targeting. The acetyl transferase p300 was capable of acetylating theunacetylated peptide and the mono acetylated peptides without any additional auxiliaryproteins. Comparing the amount of transferred acetyl groups revealed that the acetylationefficiency of the second lysine is down regulated by the first acetylation event.

Summary

IV

In CHAPTER THREE expressed protein ligation (EPL), including native chemical ligation(NCL), was used to investigate the HR of ER and its PTMs in the context of their impact onthe LBD. An N’ terminal cysteine in a LBD construct, necessary for EPL, was generated bythe methionine excision approach, while the required thioester peptides were synthesized bysolid phase peptide synthesis. Ligation of the posttranslationally modified HR peptides to amodel protein could be successfully demonstrated, however, the transfer of this method toER was not successful. The major issue in this case was protein instability under ligationconditions. Studies focusing on the extension of the ER LBD in the direction of the HRrevealed that protein expression allowed the generation of ER LBD constructs including theHR up to position 251. However, binding studies identified the HR sequence before position271 to significantly decrease the ability of ER to bind coactivator peptides, thus hinting atthe existence of two regions of the HR, which differ in terms of their structure and function.CHAPTER FOUR deals with the ER LBD screening for novel natural peptide binders by

means of ribosome display. Earlier rounds of ribosomal enrichment witnessed the expectedemergence of the well studied LXXLL motif. However, subsequent rounds led to theidentification of a more mature PXLXXLLXXP consensus, which could be validated by bothbiochemical and cellular methods. MD and X ray crystallography studies revealed a specificrole for the flanking prolines as helix breakers, which prime the helix length for optimalinteraction with the surface charge clamp. Furthermore, the conformational constraintsimposed by the prolines on adjacent immediate flanking amino acid residues are believed todetermine binding affinity through the precise orientation of side chain functionality at theER surface. This work does not only represent a fundamental re evaluation of theNR coactivator interaction and thus sets new, minimal, structural parameters based on linearsequences of proteinogenic amino acids, but also provides insights for peptide derived toolsand more druggable peptide agents.MD simulation was used to predict the propensity of short peptides to fold into an helix

in CHAPTER FIVE. Based on results from ER screening (CHAPTER FOUR), peptides weredesigned in silico to potentially function as androgen receptor (AR) coactivator bindinginhibitors (CBIs). The scaffold sequence was based on a combination of a natural NRcoactivator sequence and the NTD of AR, known to specifically interact with the AR LBD viaa phenylalanine rich FXXLF motif. Proline residues were inserted into the flanking regions ofthe central binding motif to define the helix length. MD simulation studies were performedto acquire information about the propensities of the designed peptides to form helices.Based on these results the solid phase synthesis of selected peptides was performed. CDstudies of these peptides clearly confirmed the predicted helical character of the 11 and12 mers. The AR binding affinities of the peptides was evaluated by varying the length of thehelical segment of the peptides via the proline positioning and by elaborating the adjacentamino acid of the N’ terminal proline in terms of charge, size, and hydrophobicity. This

Summary

V

conceptual approach resulted in short peptides that interact exclusively with AR LBD andoverpower known AR specific binding in a cellular context with the fully intact AR. Thus,the identified sequences can be considered as lead structures for the future development ofAR selective CBIs.The retinoid X receptor (RXR) is an important transcription factor involved in the

regulation of gene networks connected to cell growth, cell differentiation, and cell death.Misregulation of RXR or pathways that are based on RXR heterodimer formation with otherNRs can lead to various malignancies, such as cancer, cardiovascular and inflammatorydiseases. Thus in CHAPTER SIX, microarray based nuclear receptor coregulator interactionprofiling was performed to screen a small compound library of biaryl natural products (NPs)for potential RXR binding. Two NPs featuring (partial) agonism functionality could beidentified, while another NP showed non competitive antagonism characteristics throughallosteric RXR binding. Both the lipophilic side chain groups and the hydroxyl functionalityof these NPs were shown to be important aspects for their activity. Further modifications tothe two lipophilic groups present in the NPs, towards an improved mimic of the structure ofleucine side chains of coactivator peptides, resulted in an improved ability to inhibit theRXR coactivator interaction in the profiling assay. This data in combination with thestructural similarity with known NR CBIs revealed the first NP CBI, also with potential foroptimization.This thesis provides molecular and structural insights to understand function and

regulation of the structural elements of NRs. This is not only important for theunderstanding of receptor pharmacology, but also provides novel entries to target theseproteins in a specific manner.

Deutsche Zusammenfassung für Nicht-Biochemiker

VI

Anwendungen in der Chemischen Biologie für KernrezeptorenMolekulare und Strukturelle Einblicke

Kernrezeptoren sind eine Klasse von Proteinen, die sich hauptsächlich im Zellkern vonvielzelligen Tieren befinden. Proteine von dieser Proteinfamilie besitzen sowohl Bereiche mithoher Ähnlichkeit der Aminosäurenabfolge als auch Bereiche, die verhältnismäßigunterschiedlich sind. Dennoch ist die Gesamtstruktur der verschiedenen Rezeptoren sehrähnlich. Ihre Aktivität wird durch kleine, fettlösliche Moleküle reguliert. Folge der Bindungdieser Moleküle ist eine Veränderung der Rezeptorkonformation, die es wiederum anderenProteinen sogenannten Kofaktoren ermöglicht, ebenfalls mit dem Rezeptor inWechselwirkung zu treten. Eine weitere Besonderheit der Kernrezeptoren ist die Fähigkeit,an bestimmte Bereiche der DNA zu binden. Auf diese Weise kontrollieren sie dieHerstellung anderer Proteine, die für unterschiedlichste Abläufe im Körper zuständig sind.Fehler in diesem empfindlichen System können zur Entstehung von diversen Krankheitenführen, wie etwa Brustkrebs im Falle einer Fehlregulierung des Estrogenrezeptors.Medikamente konkurrieren oft mit dem kleinen, fettlöslichen Molekül um dieBindungstasche im Rezeptor, unterdrücken damit dessen Fähigkeit Kofaktoren zu bindenund vermindern somit die oft unkontrollierte Produktion von Proteinen, die die Krankheithervorgerufen haben. Eine alternative Herangehensweise ist das Design von Wirkstoffen, dieden bereits aktivierten Rezeptor direkt an dem Bereich binden, der normalerweise dieSchnittstelle zwischen Rezeptor und Kofaktoren bildet. Die Folgen sind ähnlich wie bei derzuerst genannten Variante, jedoch ist sie weniger anfällig für die Ausbildung vonResistenzen gegen Wirkstoffe ein bekanntes Problem in der Medikamentenforschung.Die Stabilität von Kernrezeptoren und die damit zusammenhängende Aktivität stehen im

direkten Zusammenhang mit Modifikationen von Aminosäuren dieser Proteine. DieseVeränderungen in Form von Hinzufügung verschiedenster Molekülgruppen haben einenentscheidenden Einfluss auf die Struktur des Rezeptors, im Speziellen in Bereichen dienormalerweise wenig strukturiert sind. Somit können Modifikationen dieser Art dieFunktionalität einzelner Rezeptorbereiche regulieren, die wiederum die Gesamtaktivität desRezeptors beeinflussen. In KAPITEL ZWEI wurde eine natürlich vorkommende Modifikation(acetylation) der Aminosäure Lysin und ihr Einfluss auf den relativ unstrukturierten Bereich(hinge region, HR) des Estrogenrezeptors untersucht. Computersimulationen (MD) zeigtenstrukturelle Veränderungen als Folge der Lysinmodifikation, wobei doppelte Modifikationden größten Effekt hatte spiralförmige Strukturen ( helix) auszubilden. Die Herstellung(SPPS) der simulierten Peptide und die anschließende Strukturanalyse (NMR, CD)bestätigten diese Erkenntnisse. Die Entdeckung einer unüblichen helikalen Struktur(PPII helix) könnte wichtig sein, um den Mechanismus zu verstehen, aber auch, um gezielteBindepartner zu entwickeln. Weitere Untersuchungen in dem Kapitel beschäftigten sich mit


VII

einem Enzym (p300 acetyltransferase), das die Übertragung der Molekülgruppe auf ein Peptidunterstützen kann. Diese zeigten, dass eine Modifizierung von Peptiden basierend auf derAminosäurenabfolge des Estrogenrezeptors mit dem Enzym möglich ist. Jedoch scheint es,dass die Übertragung der zweiten Molekülgruppe von der ersten negativ beeinflusst wird.Im DRITTEN KAPITEL wurde eine Methode verwendet, die ein chemisch synthetisiertes

Peptid und ein biologisch hergestelltes Protein miteinander verbindet (expressed proteinligation, EPL). Mit dem resultierenden Hybrid soll der Einfluss des relativ unstrukturiertenBereichs (HR) und dessen Modifikationen (acetylation) auf den Teil des Rezeptors, derverantwortlich ist für die Wechselwirkungen mit kleinen Molekülen und anderen Proteinen(ligand binding domain, LBD), untersucht werden. Damit diese chemische Reaktion (nativechemical ligation, NCL) stattfinden kann muss sowohl das Peptid als auch das Proteinspezielle Charakteristika aufweisen (thioester/N‘ Cys). Das hergestellte Peptid mitLysinmodifikation konnte erfolgreich an ein Testprotein geknüpft werden, jedoch war dieÜbertragung auf Estrogenrezeptor nicht möglich. Hauptproblem war hierbei Instabilität desProteins unter den erforderlichen Reaktionsbedingungen. Studien zu einer sukzessivenVerlängerung der LBD Region in Richtung HR zeigten, dass eine Proteinherstellung für alleKonstrukte möglich war. Jedoch zeigten anschließende Bindungsstudien, dass einÜberschreiten der Position 271 (Mitte von HR) eine entscheidende Verringerung derFähigkeit Cofaktoren zu binden zur Folge hatte. Dies legitimiert die Hypothese, dass sich dieHR in zwei Abschnitte aufteilen lässt, die sich sowohl in Struktur als auch in Funktionunterscheiden.Das VIERTE KAPITEL befasst sich mit der Schnittstelle zwischen Estrogenrezeptor und

Kofaktoren. Diese Bindungsfurche wird verwendet, um neue Peptid basierte Inhibitoren zuidentifizieren. In einem sogenannten Hochdurchsatz Screening (Ribosome Display) werdenzunächst Peptide gefunden die eine bekannte Anordnung von der fettlöslichen undungeladenen Aminosäure Leucin aufweisen (LXXLL). Bei weiteren Runden werdenzusätzlich noch Proline wasserlösliche Aminosäuren mit besonderen strukturellenEigenschaften identifiziert, die zwei Positionen vor und drei Positionen nach dem Mustergehäuft auftreten (PXLXXLLXXP). Der Vorteil des Vorhandenseins dieser Aminosäurenkonnte sowohl in Bindungsstudien als auch in Strukturanalysen (MD, X ray) bestätigtwerden. Prolin unterbricht also die spiralförmige Struktur und liefert somit die idealeBindungsform im Besonderen im Bezug auf Größe um den Estrogenrezeptor zu binden.Des Weiteren wird angenommen, dass nebenliegende Aminosäuren wegen der besonderenAnordnung von Prolin eine bevorzugte Position einnehmen. Somit präsentiert diese ArbeitErkenntnisse, die bei der Entwicklung von Peptid basierten biochemischen Methoden undWirkstoffen genutzt werden können.Fehler in der Regulierung des Androgenrezeptors wird auch mit verschiedenen

Krankheiten in Verbindung gebracht, wie zum Beispiel Prostatakrebs. In KAPITEL FÜNF


VIII

wurden, basierend auf den Erkenntnissen des Estrogenrezeptors (KAPITEL VIER), Peptide alspotentielle Inhibitoren für den Androgenrezeptor entworfen. Computersimulationen zeigten,ob unterschiedliche kurze Peptide die Tendenz hatten, eine spiralförmige Konformationeinzunehmen. Hierbei basierte das Grundgerüst auf einer Aminosäurenabfolge die fastausschließlich mit dem Androgenrezeptor wechselwirkt (FXXLF). Zusätzlich wurden Prolinevorher und nachher eingefügt, um die Länge der Spirale ( helix) zu definieren. Peptide dietheoretisch (MD) die besten Strukturen aufwiesen, wurden hergestellt (SPPS) und bezüglichStruktur (CD) und Bindeeigenschaften (polarization assay) analysiert. Die Affinität zumAndrogenrezeptor wurde einerseits durch Veränderung der Prolinposition und andererseitsdurch das Austauschen der Aminosäure, die direkt nach dem ersten Prolin liegt, optimiert.Hierbei wurden positive, negative oder neutraler Aminosäuren unterschiedlichster Größenmit fettlöslichen beziehungsweise wasserlöslichen Eigenschaften verwendet. Dieser Ansatzresultierte in kurzen Peptiden die exklusiv den Androgenrezeptor binden und auch inSäugerzellen mit bekannten Kofaktoren konkurrieren können. Somit sind diese Peptideeinsetzbar als Vorlage für Wirkstoffe für Androgenrezeptor basierende Krankheiten.Der Retinoidrezeptor (retinoid X receptor) ist auch ein Kernrezeptor und spielt eine

entscheidende Rolle beim Zellwachstum, bei der Zellteilung und beim Zelltod, aber auch beider Regenration von Knochen. In KAPITEL SECHS wird ein kleiner Satz von Naturstoffen, dieaus Magnolien gewonnen wurden, mit Hilfe einer parallelen Analyse von mehrerenKofaktorbindungsnachweisen (micro array) untersucht. Aus diesen Daten kann manherausfinden, ob ein Moleküls aktivierend (agonist) oder inaktivierend (antagonist/inhibitor)auf den Retinoidrezeptor wirkt. Einige zeigten dabei aktivierende Eigenschaften, wobeiandere die Fähigkeit, Kofaktoren zu binden, komplett unterdrückten. Weitere Analysen(polarization assay) zeigten, dass einige Naturstoffmoleküle nicht, wie erwartet, mit demnatürlichen Rezeptor bindenden, fettlöslichen Molekül konkurrieren, sondern den Rezeptoran einer anderen Stelle höchstwahrscheinlich Kofaktorbindungsstelle binden und somitdas wechselwirken mit Kofaktoren verhindern. Veränderungen in den Seitenketten derNaturstoffe, die die Struktur der Kofaktoren nachahmen (LXXLL), haben eine verbesserteAffinität zum Rezeptor zur Folge und stärken die Hypothese, dass die Bindung tatsächlichanstelle des Kofaktors stattfindet. Damit bringt diese Studie den ersten Naturstoff hervor, deran die Kofaktorbindungsstelle von Kernrezeptoren bindet.Diese Doktorarbeit liefert molekulare und strukturelle Einblicke, die für das Verständnis

von Funktion und Regulation von strukturellen Elementen von Kernrezeptoren wichtig sind.Dieses Wissen ist Voraussetzung, neue Wirkstoffe entwickeln zu können, die ein speziellesZiel nicht nur stark, sondern auch exklusiv binden, um Nebenwirkungen zu verhindern.

List of Publications

IX

List of Publications

Mocklinghoff, S.; van Otterlo, W. A. L.; Rose, R.; Fuchs, S.; Zimmermann, T. J.; Dominguez

Seoane, M.; Waldmann, H.; Ottmann, C.; Brunsveld, L., Fragment Based Design of Novel

Estrogen Receptor Tetrahydroisoquinoline Ligands, Journal of Medicinal Chemistry 2011, 54,

2005–2011.

(to be) submitted

Fuchs, S.; Nguyen H. D.; Burton M. F.; Nieto L.; de Vries van Leeuwen I. J.; Phan T. T. P.;

Schmidt A.; Agten, S. M.; Rose R.; Ottmann, C.; Milroy L. G.; Brunsveld L., Proline primed

helix length for optimal regulation of nuclear receptor coactivator protein binding, submitted.

Milroy, L. G.; Fuchs, S.; Scheepstra, M.; Lam, C. V.; Hirsch, A. K. H.; In het Panhuis, L.;

Brunsveld, L., A natural product inspired three way switch in nuclear receptor activity, to be

submitted.

Fuchs, S.; Nieto, L.; de Vries van Leeuwen I. J.; Brunsveld, L., In silico design of androgen

receptor coactivator binding inhibitors, to be submitted.

Curriculum Vitae

X

Curriculum Vitae

Sascha Fuchs was born on February 16th, 1984 in Gütersloh,Germany. After secondary education at the Werdener Gymnasiumin Essen he started studying Chemical Biology at the TUDortmund University (both Germany). He performed part of hismaster’s program in the Department of Biochemistry andMolecular Biology at the University College in London (UnitedKingdome). During his research project in the group Prof. Dr.Herbert Waldmann at the Max Planck Institute of MolecularPhysiology in Dortmund in cooperation with Prof. Dr. Henning D.

Mootz at the TU Dortmund University (both Germany), he investigated the in vivomodulation of estrogen receptor activity by conditional protein splicing. In December 2008he joined the Chemical Biology group in the Department of Biomedical Engineering at theEindhoven University of Technology (The Netherlands) starting his PhD position with Prof.Dr. Ir. Luc Brunsveld. His research focused on gaining molecular and structural insights todevelop deeper understanding of the function and regulation of the structural elements ofnuclear receptors, including estrogen , androgen and retinoid X receptor. His findings wereimportant for the understanding of receptor pharmacology and provided novel entries totarget these proteins. In this thesis the most important results are summarized.

Acknowledgements

XI

Acknowledgements

Now, I come to the end of my thesis. But not before thanking some people that had a direct orindirect influence on my life and my work over the last four years in Eindhoven, thus beingeventually co responsible for the thesis in its final version.First of all, I want to thank Prof. Dr. Ir. Luc Brunsveld. Luc, thank you for giving me the

opportunity to work on this interesting and challenging topic. I felt always welcome to enter youroffice and discuss the project if you had time (and even if you had not). I remember thatsometimes even I had to remind you about subsequent appointments as we both lost ourselves inthe depths of intense discussion. Thank you for all the fruitful discussions about differentscientific topics over the last years and your many suggestions and advice in several moments. Itwas also nice to be part of a growing, interdisciplinary group. I learned a heck of a lot these lastfew years.I want to express my gratitude to the members of my defense committee Prof. Dr. Constant

A.A. van Boeckel (TU/e & University Leiden), Dr. Christian Ottmann (TU/e), Univ. Prof. Dr.Christian F. W. Becker (University Vienna) and Prof. Dr. Carlie J.M. de Vries (University ofAmsterdam) for the critical reading of my thesis and for making suggestions to improve on itsquality. Furthermore, I want to thank Dr. Maarten Merkx and Dr. Lech Gustav Milroy forcompleting the committee. Lech, I want to thank you additionally for all the help especially in thefinal stage of my PhD. I can remember funny peptide synthesis days, excited discussions duringthe writing phase of ‘the paper’ and critical but always fair comments during the correction phaseof my thesis. You taught me a lot about scientific writing and without you and your suggestions Iwould not be able to deliver the thesis in its present state. Thanks a lot for that.Additionally I want to thank different people for their collaboration and support during my

thesis. Dr. Lidia Nieto: Lidia, thank you a lot for all the help with the structural analysis of thepeptides, namely molecular simulations, circular dichroism and NMR measurements (CHAPTER

TWO, FOUR & FIVE). Your knowledge certainly improved the quality of my thesis. ¡Gracias! Dr.Ingrid J. de Vries van Leeuwen: Ingrid, thank you for all the cellular assays you performed toevaluate the peptides (CHAPTER FOUR & FIVE). You were always willing to help and it was nice towork together with you. Dank je wel! Dr. Matthew Burton and Stijn M. Agten: Matt and Stijn, Iwant to thank you for all the work you made on the proline peptides (CHAPTER FOUR). Matt, youI want to additionally thank for all other help and discussions during the middle phase of myPhD. I enjoyed working together with you. Thanks! Dr. Hoang D. Nguyen, Dr. Trang T.P. Phanand Parisa M. Goodarzifard: Hoang, Trang and Parisa, I want to thank you for all the ribosomedisplay work you have performed (CHAPTER FOUR). C m n b n! Parisa, you I want to thank alsofor the work you did during your master’s thesis. You learned quick and worked always veryindependently. Remco Arts: Mijn koning, thank you for all the work you did on the Hinge Regionproject (CHAPTER TWO), but also for joining me to celebrate the BVB. Chan Vinh Lam: Chan Vinh,I want to thank you for thework you did during your stage (CHAPTER SIX). You were always

Acknowledgements

XII

interested in understanding the big picture. Thank you, my students! Dr. Christian Ottmann, Dr.Sabine Möcklinghoff and Andrea Schmidt: Christian, Sabine and Andrea, thank you for all yourhelp during protein crystallization and with solving the structure. Crystal fishing and patienceare definitely highly connected. I also want to thank the different OGO students and WouterEngelen and Maarten Bakker that were part of the androgen receptor project. I liked yourenthusiasm and your research formed the basis for a lot of the work presented in CHAPTER FIVE. Ialso wish to thank the analytical lab with Joost L.J. van Dongen, Ralph A.A. Bovee and XianwenLou. Your accurate measurements were on different occasions undoubtedly necessary.Furthermore I want to thank the people that were not scientifically working together with me,

but no less important for my thesis. Only a nice environment makes it possible to workoccasionally more than a standard eight hour day. Thus, for me it is very important to ‘turn’some of my colleagues into friends, as some of you might have experienced. In the beginningwere all these people very important that had made the German Dutch transition just somemonth before me. Dana, Katja, Maëlle und Dung, you prepared everything in Eindhoven for mybelated arrival. Due to you I quickly felt like home in Eindhoven. Although I left you office wisesome time later, you were very important during my time in Eindhoven, but also during mymaster’s in Dortmund. Maëlle, I enjoyed working together with you in the first few months andwe always had fun together. Merci beaucoup! Dung, we had a lot funny conversations about allkinds of topics, including football and ManU. C m n b n! Dana, du warst zwar nicht immermeiner Meinung, hast diese aber immer respektiert und ich fand unsere Gespräche immerinteressant. Katja, oder besser gesagt Mullik, du hast mich als einzige (neben Luc) von Anfang bisEnde begleitet. Als ich Anfang 2008 in die Gruppe gekommen bin noch als alleinerziehendeMullik, dann später in Eindhoven dann aber zusammen mit Pullik. Ihr seid gerade in derEndphase meiner Thesis wirklich fast wie Eltern für mich geworden. Ihr habt euch um michgesorgt, ihr habt mir Essen gebracht und mich zur Vernunft gemahnt. Diese Kombination kannteich zuvor nur von meinen Eltern. , auch dafür, dass du mein Paranimfe bist. But duringmy four years there were numerous other people responsible for making my time at work soenjoyable. Melissa, thank you for both your scientific help and also for being there for otherthings especially in the early phase of my PhD. Thanks! Ewelina, you were also importantespecially in the third year. Our coffee breaks, the cooking but also your critical questions aboutthe state of my research and the pressure you generated motivated me to work efficiently and tofinalize parts of my projects. Dzi kujemy! Micha, auch wenn nicht der identische Charakterhaben wir uns immer gut verstanden. Wir hatten immer jehörich Spaß, ob beim Fußball oder inLondon, auch wenn du dich des Öfteren zu Tode erschrocken hast wenn ich dich angesprochenhabe. Pauline, the coffee/water breaks were always needed especially during my writing phase.Merci beaucoup, also for the very last proof reading.But also all the other members of the group I would like to thank. Wencke, Inga, Ralph,

Chritian and all the students. Also I want to thank Abidin for all the support especially in the lastyear and for being my paranimfe. Also thank you for ‘sharing’ your friends, like Ba ar, Seda,Gökhan, Kamil and Alessandra. The barbecue in Breda, the football game Germany Italy in Essen

Acknowledgements

XIII

(although we lost), the short meeting in Delirium in Brussels, the gypsy festival and theChristmas market in Dortmund were all so much fun! Te ekkürler and hope to meet you all inIstanbul. I also want to thank Daniele, Mindaugas, Benjamin, Brian, Luuk, Asish, Lorenzo, Elisa,Miguel and all the others for joining in the Champions League viewing in the Irish, food in DeBengel or Touch of India, and diverse parties and other activities in Eindhoven.I also want to thank my office members: Mantas (Malisauskas), Lars (Röglin), Luc (Scheres),

Ingrid (de Vries van Leeuwen) and the current members Laurens, Parisa and Chan Vinh. Iremember funny table football competitions and Lars, der Spruch des Monats hängt hier immernoch an der Wand ‚Es ist so warm heute! Ich habe mir überlegt… !‘. Furthermore, I want to thankall the members of the bio lab and the whole cell lab family. Thanks for all the scientificdiscussions and, Björne, I hope the cell lab family tree will continue growing. Also thank you toall the members of MST/SMO. I enjoyed the interdisciplinary environment and also the socialevents including uitje and Nolte Mijer Cup, although we won football only in the year I did notjoin. I also want to thank my housemates: Oliver, Sapo and Št pán. Danke, grazie, d kujeme, youfor being so uncomplicated and helpful. I think all of you are very easy to live with. That is whatyou can tell your girl friend or future partner. But I also want to thank people I met onconferences and meeting and are still in touch with me: Annie and Christian, Marije, Randy,Ulrike, James, tack, dank je wel, thanks, danke for the nice time in Dubrovnik, Spetses,Amsterdam and all kinds of support (including Movember) and scientific exchange.Auch möchte ich mich bei meinen Freunden aus Deutschland bedanken. Zunächst Julian und

Florian für das Korrekturlesen verschiedener Teile der Arbeit. Danke für das zum Teil sehrspontane zur Verfügung stehen. Auch möchte ich Tesi und Flori danken für die Unterstützungjeglicher Art in den letzten Jahren. Ihr habt es mir leicht gemacht von den oft stressigenAbschnitten zu erholen. Aber auch meinen vielen anderen Freunden Andy, Steffi, Nadine, Yeti,Ricci, Philipp, Sandra, Pascal, Andi, Nadine, Ahmad, Katrin, Tobi, Marius, Christian, Jakob,Hendrik, Alexander, Linda, Dominik, Emily, Björn, Julia, Toby, Hanna, Claudia, Thomas,Christina, Mehrina, Sujal, Jasmine, Vera, Olaf, um nur einige zu nennen (ich habe auf jeden Fallwelche vergessen).Auch möchte ich meiner Familie danken. Franca, Mama und Papa, ich danke euch nicht nur für

die bedingungslose Unterstützung in den letzten vier Jahren, sondern auch für die Ruhe undGeduld in den letzten 27 Jahren. Ohne euch hätte ich die Doktorarbeit, das Studium, oder auchdie Schule erst gar nicht geschafft. Ich werde euch das nie vergessen und kann euch gar nichtgenug dafür danken. Klaus und Christian, auch euch danke ich für all die Unterstützung obdirekter oder indirekter Art. Ebenfalls ein großes Dankeschön geht an meine zukünftigenSchwiegereltern, Schwager und Schwägerin. Udo, Dagmar, Laura, Gio, Victor, Annie und Ferdi,ihr habt mich herzlich in eure Familie aufgenommen und ich fühle mich fast wie ein Sohn oderBruder. Danke für alles!Last but not least, möchte ich mich natürlich bei Sarah bedanken. Sarah, thank you for all your

support. Du hast immer an mich geglaubt and you always motivated me. Du bist der wichtigsteMensch in meinem Leben. Dankie!

Sascha

Chemical biology approaches for nuclear receptors : molecular … · • The final author version and the galley proof are versions of the publication after peer review. ... 4.2.

Documents